2548

Diabetes Care Volume 37, September 2014

EPIDEMIOLOGY/HEALTH SERVICES RESEARCH

Impact of an Intensive Lifestyle

Intervention on Use and Cost
of Medical Services Among
Overweight and Obese Adults
With Type 2 Diabetes: The Action
for Health in Diabetes
Diabetes Care 2014;37:25482556 | DOI: 10.2337/dc14-0093

OBJECTIVE

To assess the relative impact of an intensive lifestyle intervention (ILI) on use and
costs of health care within the Look AHEAD trial.

Mark A. Espeland,1 Henry A. Glick,2

Alain Bertoni,3 Frederick L. Brancati,4,
George A. Bray,5 Jeanne M. Clark,4
Jeffrey M. Curtis,6,7 Caitlin Egan,8
Mary Evans,9 John P. Foreyt,10
Siran Ghazarian,11 Edward W. Gregg,12
Helen P. Hazuda,13 James O. Hill,14
Don Hire,1 Edward S. Horton,15
Van S. Hubbard,9 John M. Jakicic,16
Robert W. Jeffery,17 Karen C. Johnson,18
Steven E. Kahn,19 Tina Killean,6,7
Abbas E. Kitabchi,18 William C. Knowler,6,7
Andrea Kriska,16 Cora E. Lewis,20
Marsha Miller,14 Maria G. Montez,13
Anne Murillo,19 David M. Nathan,21
Ebenezer Nyenwe,18 Jennifer Patricio,22
Anne L. Peters,11 Xavier Pi-Sunyer,22
Henry Pownall,10 J. Bruce Redmon,17
Julia Rushing,1 Donna H. Ryan,5
Monika Safford,20 Adam G. Tsai,23
Thomas A. Wadden,2 Rena R. Wing,8
Susan Z. Yanovski,9 and Ping Zhang,12 for
the Look AHEAD Research Group

RESEARCH DESIGN AND METHODS

A total of 5,121 overweight or obese adults with type 2 diabetes were randomly
assigned to an ILI that promoted weight loss or to a comparison condition of
diabetes support and education (DSE). Use and costs of health-care services were
recorded across an average of 10 years.

Department of Biostatistical Sciences, Wake

Forest School of Medicine, Winston-Salem, NC
2
Weight and Eating Disorder Program, University of Pennsylvania, Philadelphia, PA
3
Department of Epidemiology and Prevention,
Wake Forest School of Medicine, Winston-Salem,
NC
4
Johns Hopkins School of Medicine, Baltimore,
MD
5
Pennington Biomedical Research Center, Baton
Rouge, LA
6
Southwest American Indian Center, National
Institute of Diabetes and Digestive and Kidney
Diseases, Phoenix, AZ
7
Southwest American Indian Center, National
Institute of Diabetes and Digestive and Kidney
Diseases, Shiprock, NM
8
Weight Control and Diabetes Research Center,
Brown Medical School/The Miriam Hospital,
Providence, RI
9
National Institute of Diabetes and Digestive and
Kidney Diseases, National Institutes of Health,
Bethesda, MD

10
Department of Medicine, Baylor College of
Medicine, Houston, TX
11
Roybal Comprehensive Health Center, Los
Angeles, CA
12
Centers for Disease Control and Prevention,
Atlanta, GA
13
University of Texas Health Science Center at
San Antonio, San Antonio, TX
14
Anschutz Health and Wellness Center, University
of Colorado Health Sciences Center, Aurora, CO
15
Department of Clinical Epidemiology, Joslin
Diabetes Center, Boston, MA
16
Diabetes Unit, Department of Health and Physical Activity, University of Pittsburgh, Pittsburgh, PA
17
Division of Epidemiology and Community
Health, University of Minnesota, Minneapolis,
MN
18
Department of Preventive Medicine, University
of Tennessee Health Sciences Center, Memphis, TN
19
Department of Medicine, University of Washington, Seattle, WA
20
Preventive Medicine, University of Alabama at
Birmingham, Birmingham, AL

21
Massachusetts General Hospital, Boston,
MA
22
Division of and Department of Medicine, St.
Lukes-Roosevelt Hospital, New York, NY
23
Division of Internal Medicine, University
of Colorado Health Sciences Center, Aurora, CO

Corresponding author: Mark A. Espeland,

mespelan@wakehealth.edu.
Received 12 January 2014 and accepted 6 March
2014.
Clinicaltrialreg.no.NCT00017953, clinicaltrials.gov.
This article contains Supplementary Data online
at http://care.diabetesjournals.org/lookup/
suppl/doi:10.2337/dc14-0093/-/DC1.

Deceased.

2014 by the American Diabetes Association.

Readers may use this article as long as the work
is properly cited, the use is educational and not
for prot, and the work is not altered.

See accompanying articles, pp. 2424,

2442, and 2557.

care.diabetesjournals.org

RESULTS

ILI led to reductions in annual hospitalizations (11%, P = 0.004), hospital days

(15%, P = 0.01), and number of medications (6%, P < 0.001), resulting in cost
savings for hospitalization (10%, P =
0.04) and medication (7%, P < 0.001).
ILI produced a mean relative per-person
10-year cost savings of $5,280 (95% CI
3,3857,175); however, these were not
evident among individuals with a history of cardiovascular disease.
CONCLUSIONS

Compared with DSE over 10 years, ILI

participants had fewer hospitalizations, fewer medications, and lower
health-care costs.
The number of adults who both are
overweight or obese and have type 2
diabetes is increasing rapidly (1). In the
U.S., .35% of adults are obese (2), and
the projected number of diabetes cases
exceeds 30 million (3). Separately, obesity and diabetes markedly increase
health-care costs (46), and their coexistence further increases costs (7). Strategies to reduce the economic impact of
these trends are needed.
Lifestyle interventions aimed at promoting long-term weight loss and
increased physical activity are recommended for overweight and obese individuals with type 2 diabetes. Although
reduced health-care costs have been
suggested to accompany these interventions (814), no studies have prospectively recorded their long-term
effects on health-care costs among
these individuals. The Action for Health
in Diabetes (Look AHEAD) study is the
rst randomized clinical trial with sufcient size and duration to test whether
behavioral intervention targeting
weight loss and increased physical activity inuences long-term health-care service use and costs.
RESEARCH DESIGN AND METHODS

The design and methods of the Look

AHEAD trial have been published previously (15), as have its CONSORT (Consolidated Standards of Reporting Trials)
diagram and the results for its primary
outcome (16). Look AHEAD was a singleblinded randomized controlled trial that
recruited 5,145 individuals (from 2001
to 2004) who were overweight or obese

Espeland and Associates

and had type 2 diabetes. To be eligible,

they had to meet the following criteria:
4576 years of age, BMI .25 kg/m2
(.27 kg/m2 if treated with insulin), glycated hemoglobin (HbA 1c) ,11% (97
mmol/mol), blood pressure ,160/
,100 mmHg, triglyceride level ,600
mg/dL, and successful completion of a
maximum graded exercise test. Participants at the trials 16 sites were randomly assigned with equal probability
to an intensive lifestyle intervention
(ILI) or diabetes support and education
(DSE) comparator. All participants provided informed consent. Local institutional review boards approved the
protocols.
On 14 September 2012, the study was
directed by its sponsor (National Institute of Diabetes and Digestive and
Kidney Diseases) to terminate interventions based on recommendations from
the trials data and safety monitoring
board. This recommendation was based
on an evaluation of statistical futility for
the trials primary end point, a composite of death from cardiovascular disease, nonfatal myocardial infarction,
nonfatal stroke, and hospitalization for
angina (16).
Interventions

ILI participants were assigned calorie, dietary fat, and physical activity goals (17).
Trained interventionists provided instruction and encouragement in faceto-face group and individual meetings
weekly for 6 months and three times
per month for the next 6 months. Thereafter, ILI participants were offered an
individual and group meeting each
month and periodic refresher group
meetings. DSE participants were invited
to three group sessions on general education about diabetes self-care per year
during the rst 4 years and one per year
thereafter (18). Look AHEAD investigators did not manage any medical care or
medical service use (e.g., admit participants to the hospital, change dosages of
or stop prescription medications, or prescribe rehabilitation). This management
remained in the hands of the participants health-care providers. The only
exceptions were for temporary changes
in glucose medications made by study
staff to reduce the risk of hypoglycemia
and the prescription of orlistat to 684
(27%) ILI participants, which was largely
discontinued in 2008. The median

duration of orlistat use among these

684 participants was 0.98 (interquartile
range 0.431.66) years.
Assessments and Outcomes

At baseline, demographic data, medical

history, and sources of medical care
were collected by self-report. Weight
and height were measured in duplicate
using a digital scale and stadiometer.
Hypertension was determined based
on the use of antihypertensive medications or measured blood pressure. History of cardiovascular disease was
dened by self-report of prior myocardial
infarction, stroke, coronary or lowerextremity angioplasty, carotid endarterectomy, or coronary bypass surgery.
Hospitalizations, outpatient visits (ofce, hospital clinic, or other), outpatient
tests and procedures, rehabilitation/
long-term care, and home care were assessed annually through face-to-face interviews at clinic visits and at 6-month
intervals by telephone. A validation
study conrmed that information on
hospitalizations for procedures unrelated to major study end points that
had lengths of stay of #3 days could
be collected by self-report. These accounted for 14% of hospitalizations.
For the other 86% of hospitalizations,
hospital records were reviewed for admission and discharge dates, reasons for
hospitalization, and discharge status.
Hospitalizations were categorized using
the Clinical Classications Software system from the Agency for Healthcare Research and Quality (www.hcup-us.ahrq
.gov/toolssoftware/ccs/ccsfactsheet
.jsp). Participants brought prescription
medications to annual clinic visits for recording. Follow-up was censored at
each participants last interview and,
thus, excludes health-care use and costs
associated with death. All data were
collected by centrally trained staff
who were masked to intervention
assignment.
Regression models that included age,
sex, discharge location, primary diagnosis, primary procedure, and length of
stay as predictors were tted to data
from the Nationwide Inpatient Sample
(19) to project hospitalization costs. Outpatient care costs were based on the
Medicare Physician Fee Schedule (www
.cms.gov/apps/physican-fee-schedule).
Rehabilitation, long-term care, and
home health services costs were based

2549

2550

Lifestyle Intervention Reduces Health-Care Cost

on Medicare Skilled Nursing Facility Prospective Payment System and National

Home Health Utilization statistics for
Medicare Parts A and B (www.cms
.gov). Medication costs were based on
adjusted average wholesale prices obtained from the Red Book (www.redbook.com/redbook) from January to
May 2013. Medications were grouped
by using this resource to disaggregate
the broader U.S. Food and Drug Administration classications (www.fda.gov/
drugs/developmentapprovalprocess/
howdrugsaredevelopedandapproved/
approvalapplications/investigational
newdrugindapplication/ucm176533
.htm). Costs were obtained for both
brand-name and generic-brand drugs
and assigned to medications proportional to the reported use (i.e., brand
name vs. generic) by participants. Medication doses were not recorded; the
most commonly prescribed doses were
assumed. Adjustments accounted for the
relationship between average wholesale
prices and the cost of both the manufacturing and the purchasing of drugs
(20,21). All costs were expressed in
2012 dollars, with adjustments made using the medical care component of the
Consumer Price Index (www.bls.gov/
data/ination_calculator.htm).
Statistical Analysis

We report two types of outcomes: perparticipant average annual rates and

costs of medical service use and perparticipant 10-year cumulative mean
discounted costs. Data were collected
from contacts with participants occurring before 14 September 2012. Hospitalizations reported later but that had
admission dates before 14 September
2012 were also included.
Analyses followed intention-to-treat
principles and used all available data,
with participants included in their randomization group independent of adherence. For per-participant average
annual estimates, frequencies and costs
were tallied for each participant and divided by follow-up time to obtain observed counts and costs per year. For
inference, weighted ANCOVA was used
to compare intervention groups, with
analytical weights proportional to participants lengths of follow-up. Clinic,
the sole stratication factor in randomization, was a covariate. To accumulate
costs over 10 years, annual estimates

Diabetes Care Volume 37, September 2014

were discounted at 3% per year and

summed. Bootstrapping was used for
CIs of the 10-year differences in accumulated mean costs.
The Look AHEAD protocol prespecied subgroup comparisons for the primary outcome of major cardiovascular
events were based on sex, history of cardiovascular disease, and race/ethnicity.
Although this prespecication did not
extend to other study outcomes, we report parallel comparisons for use rates
and costs. We also report results for subgroups based on age and baseline BMI.
Tests of interaction were used to assess
the consistency of differences between
intervention groups for each of these
subgroups.
RESULTS

Follow-up cost data were available from

5,121 of the 5,145 Look AHEAD participants (99.5%). Twenty-four participants
withdrew from the trial or were lost to
follow-up before the collection of any
cost data. Collectively, these 5,121 individuals provided 50,498 person-years of
follow-up (averaging 9.9 years per participant). Follow-up was terminated for
refusal and lost contact in 6.2% of ILI and
6.5% of DSE participants (P = 0.62) and
for death in 6.8% of ILI and 7.8% of DSE
participants (P = 0.15).
Baseline Characteristics

At baseline (Table 1), 25% of participants were aged 4554 years, 55%
were aged 5564 years, and 20% were
aged 6576 years. Fifteen percent were
overweight (BMI 25.029.9 kg/m2), and
22% had class III obesity (BMI $40.0
kg/m 2 ). Sixty percent were female;
54% had diabetes for at least 5 years;
14% had a history of cardiovascular disease; and 63% were non-Hispanic
white. None of the characteristics in
Table 1 differed signicantly between
intervention groups.
Service Use

ILI participants averaged 0.177 hospitalizations per year compared with 0.199
hospitalizations per year for DSE participants, an 11% (P = 0.004) reduction
(Table 2). Reductions in hospitalization
rates reached nominal statistical signicance for cardiovascular (11%, P = 0.04),
pulmonary (27%, P = 0.05), and other
(i.e., not among the named categories)
(8%, P = 0.05) diseases. There was a signicant 15% (P = 0.01) reduction in the

average annual days in the hospital (0.69

vs. 0.81 days/year for ILI and DSE participants, respectively). ILI participants also
had relative reductions of 14% (P = 0.05)
and 16% (P = 0.001) in the annual rates of
rehabilitation/long-term care and home
care, respectively. There was no signicant difference between the groups in
the rates of outpatient visits, but the ILI
group had lower rates for outpatient rehabilitation facilities, home care, and
hospitalization.
On average across follow-up, ILI participants attending clinic visits were
taking 4.65 prescription medications
compared with 4.96 medications in
DSE, a reduction of 6% (P , 0.0001).
The most signicant reductions occurred
for diabetes drugs (14%, P , 0.0001),
lipid-lowering drugs (5%, P = 0.002),
and antihypertensives (4%, P = 0.02).
Use of psychiatric/neurologic drugs
(the majority of which were agents
that may be prescribed to treat depression, neuropathy, and insomnia) was
10% (P = 0.02) greater among ILI than
among DSE participants.
Costs

The per-participant average annual cost

of health-care services and medications
was 7% less among ILI than among DSE
participants ($8,321 vs. $8,916 per year;
P = 0.002). This resulted from fewer hospitalizations and less medication use
(Table 2). The annual 11% fewer hospitalizations translated to 10% ($283)
lower per-participant average annual
costs (P = 0.04). The 6% lower annual
use of prescribed medications translated to a 7% ($281) lower annual medication cost (P , 0.0001). The most
signicant lower per-participant average annual medication costs were for
diabetes drugs (17%, P , 0.0001),
lipid-lowering drugs (6%, P = 0.005),
and antihypertensive drugs (6%, P =
0.01). The 10% greater use of psychiatric/
neurologic drugs for ILI participants translated to a 12% (P = 0.04) greater perparticipant average annual psychiatric/
neurologic drug cost.
Ten-year discounted cumulative costs
(Fig. 1AD) were $78,361 (DSE) vs.
$73,081 (ILI), a difference of $5,280
(95% CI $3,385$7,175). Assignment to
ILI yielded $2,567 ($954$4,180) lower
10-year discounted hospitalization costs
and $2,474 ($2,047$2,901) lower discounted medication costs. Ten-year

care.diabetesjournals.org

Espeland and Associates

Table 1Characteristics at the time of enrollment into the Look AHEAD trial by
intervention assignment
Characteristic
DSE (n = 2,563)
ILI (n = 2,558)
P value
Age
4554 years
5564 years
6576 years

23
55
21

25
56
19

BMI
25.029.9 kg/m2
30.034.9 kg/m2
35.039.9 kg/m2
.40 kg/m2

14
35
29
22

16
36
26
22

60
40

60
40

86
14

86
14

Race/ethnicity
African American
American Indian
Hispanic/Latino
Non-Hispanic white
Other/mixed

16
5
13
63
3

16
5
13
63
3

Diabetes duration (n = 40 missing)

,5 years
$5 years

45
55

47
53

Education (n = 110 missing)

High school or less
Post-high school
College graduate

21
38
41

20
37
42

Hypertension
No
Yes

17
83

16
84

1.20
1.40
0.53

1.24
1.38
0.53

34
37
30

34
37
29

8
77
21
4
2
4

9
77
20
5
3
4

74
12
8
5
1

74
12
8
6
1

Sex
Female
Male
History of cardiovascular disease*
No
Yes

Prescription medications, mean number

Antihypertensive
Diabetes
Lipid lowering
Annual household income (n = 503 missing)
,$40,000
$40,000 to ,$80,000
$$80,000
Health insurance status
Uninsured
Insured (self or work)
Medicare/Medicaid
Tricare/VA/military
Indian Health Service
Other/not reported
Usual source of health care (n = 13 missing)
Private physicians ofce
Hospital clinic/outpatient
Community health center
Other health-care facility
No usual source of care

0.15

0.16

0.75

0.45

0.98

CONCLUSIONS

0.20

0.54

0.34

0.29
0.51
0.86
0.84

0.71
0.66
0.13
0.36
0.32
0.56
0.73

Data are % unless otherwise indicated. VA, Veterans Affairs. *History of cardiovascular disease
included myocardial infarction, coronary artery bypass, angioplasty/stent procedure, peripheral
vascular disease, stroke, stable angina, and class I/II heart failure. Participants were able to
indicate more than one source of health insurance. x2 test.

differences in discounted outpatient

service costs were lower ($239 [2$267
to $745]).

based on age, baseline BMI, sex, and

race/ethnicity (Table 3). Intervention effects varied according to the participants history of cardiovascular disease
at baseline for outpatient (interaction
P , 0.0001), medication (P = 0.02),
and total costs (P = 0.02) but not for
hospitalization costs (P = 0.71). For participants without a history of cardiovascular disease at baseline, ILI was
associated with lower per-participant
average annual costs of $133 (95% CI
$13$252) for outpatient care, $343
($216$470) for medications, and $801
($417$1,185) for total costs. For those
with a history of cardiovascular disease,
ILI was associated with greater perparticipant average annual outpatient
costs of $592 ($286$889).

From tests of interactions, the relative effects of ILI on average annual total
costs were similar across subgroups

The Look AHEAD intervention produced

sustained differences between intervention groups in mean weight loss
(7.9 and 2.5 percentage points at 1 and
10 years in ILI and DSE, respectively) (16)
and in physical tness (22,23). Although
not signicantly reducing overall risk of
major cardiovascular disease events
(16), the intervention improved many
measures of health, including markers
of diabetes control, blood pressure,
plasma lipid levels, sleep quality, physical function, and depression (2427).
Although these effects varied in magnitude and duration, and some were not
sustained throughout follow-up, each
could be associated with reduced
health-care needs. The medical management of the participants was left to
their own health-care providers, who
were provided with clinically relevant
information, such as annual lipid and
blood pressure measurements.
Ten years of ILI broadly reduced the
use of health-care services, including
hospitalizations, selected outpatient
services, and medications, and the total
overall costs of health care, with significant savings for hospitalization and
medication costs across many disease
indications. The 11% overall relative decrease in 10-year hospitalization costs is
less than the 17% relative reduction in
10-year inpatient care costs reported by
the Diabetes Prevention Program
among individuals initially free of diabetes (13). However, because the Look
AHEAD cohort had more comorbidities
and overall hospitalizations, the absolute

2551

2552

Lifestyle Intervention Reduces Health-Care Cost

Diabetes Care Volume 37, September 2014

Table 2Average annual rates of medical service use and average annual costs over follow-up for participants grouped by
intervention assignment
Percent decrease
Category
DSE
ILI
Difference (SE)
(increase) in use
P value*
Per-participant average annual medical service use
Hospitalizations
Bone
Cancer
Cardiovascular
Metabolism
Pulmonary
Renal
Other
Days in hospital
Outpatient services
Visits
Rehabilitation center/LTC (%)
Home care (%)
Medications
Diabetes
Lipid lowering
Antihypertensive
Other cardiovascular disease
Psychiatric/neurologic
Musculoskeletal
Other

0.199 (0.006)
0.009 (0.001)
0.013 (0.002)
0.066 (0.003)
0.010 (0.001)
0.011 (0.001)
0.004 (0.001)
0.086 (0.003)
0.81 (0.03)

0.177 (0.006)
0.008 (0.001)
0.012 (0.001)
0.059 (0.002)
0.008 (0.001)
0.008 (0.001)
0.003 (0.001)
0.079 (0.003)
0.69 (0.03)

0.023 (0.008)
0.001 (0.001)
0.0005 (0.0024)
0.008 (0.004)
0.002 (0.001)
0.003 (0.001)
0.001 (0.001)
0.008 (0.004)
0.11 (0.04)

11
11
4
11
20
27
25
8
15

0.004
0.20
0.84
0.04
0.07
0.05
0.22
0.05
0.01

11.75 (0.11)
2.9 (0.17)
7.3 (0.26)
4.96 (0.05)
1.45 (0.01)
0.65 (0.01)
1.24 (0.02)
0.11 (0.005)
0.29 (0.01)
0.20 (0.01)
1.02 (0.02)

11.50 (0.11)
2.5 (0.17)
6.1 (0.26)
4.65 (0.05)
1.25 (0.01)
0.62 (0.01)
1.19 (0.02)
0.10 (0.005)
0.32 (0.01)
0.21 (0.01)
0.97 (0.02)

0.25 (0.16)
0.4 (0.20)
1.2 (0.37)
0.31 (0.06)
0.19 (0.02)
0.04 (0.01)
0.06 (0.02)
0.01 (0.01)
20.03 (0.01)
20.003 (0.01)
0.05 (0.03)

2
14
16
6
14
5
4
9
(10)
(5)
5

0.13
0.05
0.001
,0.0001
,0.0001
0.002
0.02
0.14
0.02
0.59
0.06

Per-participant average annual medical service costs

Hospitalization
Bone
Cancer
Cardiovascular
Metabolism
Pulmonary
Renal
Other
Outpatient services
Visits
Tests and procedures
Rehabilitation center/LTC
Home care
Medications
Diabetes
Lipid lowering
Antihypertensive
Other cardiovascular disease
Psychiatric/neurologic
Musculoskeletal
Other

2,789 (96)
136 (16)
219 (31)
1,024 (54)
123 (11)
137 (19)
45 (7)
1,106 (47)
2,344 (41)
1,513 (15)
510 (7)
198 (29)
123 (10)
3,784 (44)
1,226 (18)
841 (12)
436 (7)
111 (5)
329 (13)
89 (4)
752 (19)

2,506 (96)
120 (16)
213 (31)
948 (54)
91 (11)
90 (19)
32 (7)
1,012 (46)
2,313 (40)
1,502 (15)
502 (7)
201 (29)
107 (10)
3,503 (43)
1,012 (18)
793 (12)
411 (7)
96 (5)
367 (13)
96 (4)
727 (19)

283 (136)
16 (23)
6 (43)
77 (77)
32 (16)
47 (27)
12 (10)
94 (66)
31 (57)
11 (22)
8 (10)
23 (41)
16 (13)
281 (61)
214 (25)
48 (17)
25 (10)
15 (8)
238 (19)
27 (6)
25 (27)

10
12
3
3
8
26
27
8
1
1
2
(2)
13
7
17
6
6
14
(12)
(8)
3

0.04
0.49
0.89
0.32
0.04
0.08
0.22
0.16
0.59
0.61
0.48
0.93
0.24
,0.0001
,0.0001
0.005
0.01
0.06
0.04
0.23
0.35

Total annual cost

8,916 (133)

8,321 (133)

595 (188)

0.002

Data are mean (SE) unless otherwise indicated. LTC, long-term care. *Weighted ANCOVA. Metabolism includes hospitalizations for nondiabetic
endocrine disorders (e.g., thyroid); uid, electrolyte, and nutrition disorders; and other metabolic disorders (e.g., gout).

savings associated with its intervention

were greater ($2,567 vs. $1,309 [converted to 2012 dollars] for the Look
AHEAD ILI vs. Diabetes Prevention Program
behavioral intervention, respectively).
The signicant reduction in hospitalizations for cardiovascular disease differs from the nonsignicant nding for
the primary end point of Look AHEAD: a
composite of the rst occurrence of cardiovascular disease death, nonfatal
myocardial infarction, nonfatal stroke,

or hospitalization for angina (16). Unlike

the primary end point, the cardiovascular hospitalization category presented in
this article is more broadly dened and
not limited to the rst occurrence of
events.
The rate of outpatient visits did not
differ between intervention groups,
supporting the decision of the study investigators not to provide ongoing medical management to participants but to
leave this to their personal health-care

providers. Thus, the reduction in other

use and costs was not due to medical
management by study investigators of
ILI participants.
A clinical trial by Tsai et al. (14) found that
behavioral weight loss intervention in an
obese population can reduce medication
costs, which agrees with self-controlled
pre- and postobservational cohort studies of weight loss programs (28,29) and
trials that have been analyzed as preand postobservational cohort studies

care.diabetesjournals.org

Espeland and Associates

Figure 1Per-participant average 10-year cumulative discounted costs. A: Ten-year cumulative total costs. B: Ten-year cumulative medication costs.
C: Ten-year cumulative hospitalization costs. D: Ten-year cumulative outpatient care costs. Costs are reported in 2012 U.S. dollars and discounted at
3% per year.

(8,30). The Diabetes Prevention Program

reported a 6% relative reduction in
10-year medication costs among participants assigned to its lifestyle intervention (13), resulting in a relative savings of
$501 (converted to 2012 dollars). The 7%
relative reduction produced by the Look
AHEAD intervention yielded a savings of
$2,474 (in 2012 dollars) because of the
overall greater use of medications in its
cohort. The nding of a greater use of
psychiatric/neurologic medications
over time among ILI compared with
DSE participants follows a chance imbalance for greater antidepressant use at
baseline among participants randomly
assigned to ILI (17.4%) compared with
DSE (15.1%; P = 0.02). Covariate adjustment for baseline antidepressant use
eliminated differences between intervention groups in the use of these
medications during follow-up (P =
0.63).
The savings in total medical care
costs associated with ILI accrued gradually over time, resulting in a 10-year

difference of $5,280. Similar patterns

were seen for the accumulation of
both hospital and medications costs,
with 10-year differences of $2,567 and
$2,474, respectively. There were only
small overall differences during the rst
4 years of follow-up when the interventions were most intense, although diabetes, hypertension, and hyperlipidemia
medication costs were signicantly
lower among ILI participants even during the rst year after randomization
(31).
Health-care costs associated with
obesity and diabetes accelerate in later
life (32,33), and as expected, the rate at
which costs accumulated in the Look
AHEAD cohort increased with time. Future follow-up of the Look AHEAD
cohort will assess whether the intervention has a legacy effect during the
oldest years of life when health-care
costs are highest (i.e., whether the differences in accumulated costs between
the ILI and DSE cohorts continue to
diverge).

Based on interaction tests, the savings on overall medical costs associated

with ILI did not depend on participant
age, BMI, sex, or race/ethnicity. In contrast, intervention effects on outpatient,
medication, and total costs differed signicantly according to baseline history
of cardiovascular disease. For participants with no cardiovascular disease
history, assignment to ILI signicantly
reduced the costs of outpatient care
and medication; for those with a history,
assignment to ILI signicantly increased
outpatient costs and had no effect on
medication costs. The ILI effect on hospital costs did not differ between those
who had or did not have a history of
cardiovascular disease.
History of cardiovascular disease at
baseline was prespecied as a subgroup
for comparing intervention effects on
the primary outcome. The hazard ratio
for the intervention effect among participants with no history of cardiovascular disease for the primary outcome was
0.86 (95% CI 0.721.02) compared with

2553

2554

Lifestyle Intervention Reduces Health-Care Cost

Diabetes Care Volume 37, September 2014

Table 3Mean annual costs (in 2012 dollars) over follow-up by intervention assignment for subgroups of participants based
on baseline characteristics
Subgroup
DSE
ILI
Difference (95% CI)
DSE vs. ILI P value
Interaction P value
Age
4554 years
Hospitalization
Outpatient
Medication
Total
5564 years
Hospitalization
Outpatient
Medication
Total
6576 years
Hospitalization
Outpatient
Medication
Total
BMI
2529.9 kg/m2
Hospitalization
Outpatient
Medication
Total
30.039.9 kg/m2
Hospitalization
Outpatient
Medication
Total
.40.0 kg/m2
Hospitalization
Outpatient
Medication
Total
Sex
Female
Hospitalization
Outpatient
Medication
Total
Male
Hospitalization
Outpatient
Medication
Total
History of cardiovascular disease
No
Hospitalization
Outpatient
Medication
Total
Yes
Hospitalization
Outpatient
Medication
Total
Race/ethnicity
African American
Hospitalization
Outpatient
Medication
Total
American Indian
Hospitalization

1,674 (190)
1,902 (80)
3,439 (87)
7,014 (263)

1,876 (184)
1,928 (78)
3,372 (84)
7,176 (255)

2203 (2721 to 316)

226 (2246 to 193)
67 (2170 to 305)
2162 (2879 to 556)

0.44
0.81
0.58
0.66

0.14
0.85
0.13
0.09

2,713 (129)
2,313 (54)
3,892 (59)
8,918 (178)

2,355 (128)
2,272 (54)
3,560 (58)
8,187 (177)

358 (2 to 713)
41 (2109 to 191)
332 (170 to 495)
731 (239 to 1,223)

0.05
0.59
,0.0001
0.004

d
d
d
d

4,321 (208)
2,954 (88)
3,911 (95)
11,186 (288)

3,828 (218)
2,973 (92)
3,521 (100)
10,322 (302)

493 (297 to 1,084)

219 (2269 to 230)
390 (120 to 660)
864 (47 to 1,681)

0.10
0.88
0.005
0.04

d
d
d
d

2,125 (254)
2,085 (103)
3,476 (115)
7,686 (352)

2,250 (240)
2,219 (97)
3,084 (109)
7,552 (333)

2125 (2810 to 561)

2134 (2424 to 156)
392 (82 to 702)
133 (2818 to 1,084)

0.72
0.36
0.01
0.78

0.36
0.43
0.72
0.58

2,864 (121)
2,333 (51)
3,823 (55)
9,021 (168)

2,453 (122)
2,296 (51)
3,574 (55)
8,322 (169)

411 (74 to 748)

38 (2105 to 180)
249 (97 to 401)
698 (231 to 1,165)

0.02
0.60
0.001
0.003

d
d
d
d

3,001 (202)
2,537 (89)
3,867 (91)
9,405 (281)

2,829 (201)
2,425 (88)
3,603 (91)
8,857 (279)

172 (2387 to 731)

112 (2134 to 358)
264 (11 to 516)
548 (2228 to 1,323)

0.55
0.37
0.04
0.17

d
d
d
d

2,311 (123)
2,293 (52)
3,651 (56)
8,255 (171)

2,221 (123)
2,256 (52)
3,385 (56)
7,862 (171)

90 (2252 to 432)
37 (2108 to 182)
266 (111 to 421)
393 (282 to 867)

0.60
0.62
0.001
0.10

0.08
0.90
0.76
0.18

3,510 (151)
2,420 (64)
3,984 (69)
9,914 (210)

2,931 (151)
2,398 (64)
3,679 (68)
9,008 (209)

579 (160 to 998)

23 (2155 to 201)
305 (115 to 495)
906 (325 to 1,487)

0.007
0.80
0.002
0.002

d
d
d
d

2,407 (101)
2,295 (43)
3,667 (46)
8,369 (139)

2,082 (101)
2,162 (43)
3,324 (46)
7,568 (139)

325 (46 to 604)

133 (13 to 252)
343 (216 to 470)
801 (417 to 1,185)

0.02
0.03
,0.0001
,0.0001

0.71
,0.0001
0.02
0.02

5,373 (261)
2,674 (112)
4,569 (119)
12,615 (360)

5,191 (253)
3,266 (109)
4,635 (116)
13,093 (349)

181 (2533 to 895)

2592 (2889 to 2286)
267 (2392 to 259)
2478 (21,461 to 505)

0.62
0.0002
0.69
0.34

d
d
d
d

2,118 (242)
2,090 (101)
3,437 (107)
7,644 (332)

2,590 (241)
2,278 (101)
3,129 (107)
7,998 (331)

2473 (21,142 to 197)

2189 (2469 to 91)
308 (11 to 605)
2353 (21,271 to 565)

0.17
0.19
0.04
0.45

0.08
0.32
0.95
0.13

2,188 (431)

1,956 (424)

232 (2953 to 1,417)

0.70

Continued on p. 2555

care.diabetesjournals.org

Espeland and Associates

Table 3Continued
Subgroup
Outpatient
Medication
Total
Hispanic/Latino
Hospitalization
Outpatient
Medication
Total
Non-Hispanic white
Hospitalization
Outpatient
Medication
Total
Other/multiple races
Hospitalization
Outpatient
Medication
Total

DSE

ILI

Difference (95% CI)

DSE vs. ILI P value

Interaction P value

1,479 (180)
2,649 (191)
6,316 (591)

1,607 (177)
2,503 (188)
6,066 (581)

2128 (2624 to 367)

146 (2380 to 672)
250 (21,375 to 1,874)

0.61
0.59
0.76

d
d
d

2,276 (269)
1,742 (113)
3,031 (119)
7,048 (369)

1,468 (267)
1,545 (112)
2,654 (118)
5,666 (366)

808 (65 to 1,551)

197 (2114 to 508)
377 (47 to 707)
1,382 (363 to 2,401)

0.03
0.21
0.02
0.008

d
d
d
d

3,072 (120)
2,583 (50)
4,104 (53)
9,760 (164)

2,752 (120)
2,511 (50)
3,845 (53)
9,108 (164)

320 (213 to 652)

72 (267 to 211)
260 (112 to 407)
652 (196 to 1,107)

0.06
0.31
0.001
0.005

d
d
d
d

3,472 (569)
2,549 (238)
3,805 (252)
9,827 (780)

2,225 (539)
2,749 (225)
3,551 (239)
8,525 (738)

1,248 (2288 to 2,784)

2200 (2843 to 443)
255 (2427 to 936)
1,303 (2803 to 3,408)

0.11
0.54
0.46
0.22

d
d
d
d

Data are mean (SE) unless otherwise indicated. Included are results from ANCOVAs to assess the consistency of differences between intervention
groups across subgroups. Negative differences occurred when costs among ILI participants exceeded those for DSE participants; positive differences
occurred when costs among ILI participants were less than those for DSE participants. The DSE vs. ILI P values are for the mean difference between
intervention groups within each strata. Interaction P values indicate whether the DSE vs. ILI differences were similar among subgroups. Weighted
ANCOVA.

1.12 (0.901.42) for those with a history

of cardiovascular disease. Although this
trend did not reach nominal statistical
signicance (P = 0.06) (16), it resembles
the difference we report for outpatient
costs. Increased outpatient costs among
ILI participants with prior cardiovascular
disease were not triggered by an imbalance in cardiovascular disease events.
Annual outpatient costs were 22% (P =
0.006) higher among ILI than among DSE
participants before these events (including those individuals with no ontrial cardiovascular outcomes), which is
similar to the overall difference in costs
(Table 2). Why the intervention may affect costs differently among participants
with and without a cardiovascular disease history is not clear; however, these
results imply that the ILI is most benecial to overweight and obese individuals
with diabetes before cardiovascular disease is diagnosed.
The study had several limitations. The
Look AHEAD cohort, although geographically and demographically diverse,
comprised volunteers to a randomized
clinical trial who were required to have a
source of usual medical care; thus, the
degree to which the ndings may generalize to other populations is unclear.
We relied on self-report to identify outpatient care and the occurrence of
hospitalizations. We did not assess
health-care costs related to death. As

noted, there were slightly fewer deaths

among the ILI than among the DSE participants, and the exclusion of these
costs may have led to an underestimation of ILI benets. Follow-up was not
complete for all participants, and the
results of the service use, cost, and subgroup analyses may be biased if followup lengths were different. However, we
used a pattern-mixture approach (34) to
assess the sensitivity of the ndings,
examining the consistency of results
across strata based on length of followup, which conrmed the ndings (data
not shown). We did not report intervention costs or draw conclusions from a
cost-benet analysis; these are reserved
for future study.
In conclusion, random assignment of
overweight and obese individuals with
type 2 diabetes to 10 years of an intensive behavioral intervention that
focused on weight loss and increased
physical activity resulted in relatively
fewer hospitalizations, fewer days in
the hospital, and less use of prescription medications. Cumulatively, these
effects resulted in an average annual
savings of almost $600 per participant
relative to a comparison condition
DSE.

Duality of Interest. M.A.E. serves on monitoring boards for Terumo Medical Corporation and

the Kowa Research Institute. He serves on a

steering committee for Boehringer Ingelheim
and has recently served on an advisory committee for Takeda Global Research. H.A.G. has
twice been a sponsored lecturer by Merck. J.O.H.
serves on advisory boards for Takeda and
Novo Nordisk. E.S.H. has received consulting,
advisory board, monitoring board, and speakers
board support from Amgen; Amylin Pharmaceuticals, LLC; Bristol-Myers Squibb/AstraZeneca; GI
Dynamic, Inc.; Gilead Sciences, Inc.; International Medical Press Global Partnership for Effective
Diabetes Management; Janssen Pharmaceuticals,
Inc.; Merck Research Laboratories, Inc.; Sano,
Inc.; Vivus, Inc.; Theracos Pharmaceuticals, Inc.;
and Takeda Pharmaceuticals, Inc. E.N. has received research support from Eli Lilly and Company and GlaxoSmithKline. A.L.P. has consulted
for Abbott Diabetes Care, Becton Dickinson,
Bristol-Myers Squibb, Boehringer Ingelheim,
Janssen, Lilly, Medtronic MiniMed, and Sano;
has been on the speakers bureau for Bristol-Myers
Squibb/AstraZeneca and Novo Nordisk; has received research grant funding from Medtronic
MiniMed; and has received editorial fees from
Medscape. D.H.R. was a paid consultant/advisor
to Novo Nordisk, Janssen, Takeda, Vivus, and
Eisai and has an equity position in Scientic Intake. M.S. has received salary support from Amgen and diaDexus and has served as a consultant
for diaDexus. T.A.W. serves on advisory boards
for Novo Nordisk and Orexigen and is a consultant to Boehringer Ingelheim. No other potential
conicts of interest relevant to this article were
reported.
Author Contributions. M.A.E. and D.H. contributed to the data analysis and organization
and drafting of the manuscript. H.A.G., A.B.,
M.E., T.K., W.C.K., M.G.M., and P.Z. organized
the manuscript and wrote the initial drafts. F.L.B.,
G.A.B., J.M.Cl., J.M.Cu., C.E., J.P.F., S.G., E.W.G.,
H.P.H., J.O.H., E.S.H., V.S.H., J.M.J., R.W.J., K.C.J.,

2555

2556

Lifestyle Intervention Reduces Health-Care Cost

S.E.K., A.E.K., A.K., C.E.L., M.M., A.M., D.M.N.,

E.N., J.P., A.L.P., X.P.-S., H.P., J.B.R., D.H.R., M.S.,
T.A.W., R.R.W., and S.Z.Y. designed and conducted
the trial, collected data, and reviewed/edited the
manuscript. J.R. contributed to the data analyses.
A.G.T. contributed to the data analyses, designed and conducted the trial, collected data,
and reviewed/edited the manuscript. M.A.E.
is the guarantor of this work and, as such, had full
access to all the data in the study and takes
responsibility for the integrity of the data and the
accuracy of the data analysis.
Prior Presentation. Parts of this study were
presented in abstract form at the 73rd Scientic
Sessions of the American Diabetes Association,
Chicago, IL, 2125 June 2013.

References
1. Mokdad AH, Ford ES, Bowman BA, et al.
Prevalence of obesity, diabetes, and obesityrelated health risk factors, 2001. JAMA 2003;
289:7679
2. Ogden CL, Carroll MD, Kit BK, Flegal KM.
Prevalence of Obesity in the United States,
2009-2010. NCHS data brief no. 82. Hyattsville,
MD, National Center for Health Statistics, 2012
3. Huang ES, OGrady M, Basu A, Capretta JC.
Projecting the future diabetes population size
and related costs for the U.S. Diabetes Care
2009;32:22252229
4. Finkelstein EA, Trogdon JG, Cohen JW, Dietz
W. Annual medical spending attributable to
obesity: payer-and service-specic estimates.
Health Aff (Millwood) 2009;28:w822w831
5. Tsai AG, Williamson DF, Glick HA. Direct
medical cost of overweight and obesity in the
USA: a quantitative systematic review. Obes Rev
2011;12:5061
6. American Diabetes Association. Economic
costs of diabetes in the US in 2007. Diabetes
Care 2008;3:595615
7. Liebl A, Breitscheidel L, Nicolay C, Happich
M. Direct costs and health-related resource utilization in the 6 months after insulin initiation in
German patients with type 2 diabetes mellitus
in 2006: INSTIGATE study. Curr Med Res Opin
2008;24:23492358
8. Collins RW, Anderson JW. Medication cost
savings associated with weight loss for obese
non-insulin-dependent diabetic men and
women. Prev Med 1995;24:369374
9. Tsai AG, Glick HA, Shera D, Stern L, Samaha
FF. Cost-effectiveness of a low-carbohydrate
diet and a standard diet in severe obesity.
Obes Res 2005;13:18341840
10. Herman WH, Hoerger TJ, Brandle M, et al;
Diabetes Prevention Program Research Group.
The cost-effectiveness of lifestyle modication
or metformin in preventing type 2 diabetes in
adults with impaired glucose tolerance. Ann Intern Med 2005;142:323332

Diabetes Care Volume 37, September 2014

11. Wolf AM, Siadaty M, Yaeger B, et al. Effects

of lifestyle intervention on health care costs:
Improving Control with Activity and Nutrition
(ICAN). J Am Diet Assoc 2007;107:13651373
12. Krukowski RA, Tilford JM, Harvey-Berino J,
West DS. Comparing behavioral weight loss
modalities: incremental cost-effectiveness of
an internet-based versus an in-person condition. Obesity (Silver Spring) 2011;19:1629
1635
13. Diabetes Prevention Program Research
Group. The 10-year cost-effectiveness of lifestyle intervention or metformin for diabetes
prevention: an intent-to-treat analysis of the
DPP/DPPOS. Diabetes Care 2012;35:723730
14. Tsai AG, Wadden TA, Volger S, et al. Costeffectiveness of a primary care intervention to
treat obesity. Int J Obes (Lond) 2013;37(Suppl.
1):S31S37
15. Ryan DH, Espeland MA, Foster GD, et al;
Look AHEAD Research Group. Look AHEAD (Action for Health in Diabetes): design and methods for a clinical trial of weight loss for the
prevention of cardiovascular disease in type 2
diabetes. Control Clin Trials 2003;24:610628
16. Wing RR, Bolin P, Brancati FL, et al; Look
AHEAD Research Group. Cardiovascular effects
of intensive lifestyle intervention in type 2 diabetes. N Engl J Med 2013;369:145154
17. Wadden TA, West DS, Delahanty L, et al;
Look AHEAD Research Group. The Look AHEAD
study: a description of the lifestyle intervention
and the evidence supporting it [published correction appears in Obesity (Silver Spring) 2007;
15:1339]. Obesity (Silver Spring) 2006;14:
737752
18. Wesche-Thobaben JA. The development
and description of the comparison group in
the Look AHEAD trial. Clin Trials 2011;8:320
329
19. Healthcare Cost and Utilization Project.
Overview of the Nationwide Inpatient Sample
(NIS) Available from http://www.hcup-us.ahrq
.gov/nisoverview.jsp. Accessed 15 September
2010
20. Levinson DR. Medicaid Drug Price Comparison: Average Sales Price to Average Wholesale
Price. Publication OEI-03-05-00200. Washington, DC, Ofce of the Inspector General, Department of Health and Human Services, 2005
21. Levinson DR. Medicaid Drug Price Comparisons: Average Manufacturer Price to Published
Prices. Publication OEI-05-05-00240. Washington, DC, Ofce of the Inspector General, Department of Health and Human Services, 2005
22. Wing RR; Look AHEAD Research Group.
Long-term effects of a lifestyle intervention on
weight and cardiovascular risk factors in individuals with type 2 diabetes mellitus: four-year results of the Look AHEAD trial. Arch Intern Med
2010;170:15661575

23. Jakicic JM, Jaramillo SA, Balasubramanyam

A, et al; Look AHEAD Study Group. Effect of a
lifestyle intervention on change in cardiorespiratory tness in adults with type 2 diabetes:
results from the Look AHEAD Study. Int J Obes
(Lond) 2009;33:305316
24. Foster GD, Borradaile KE, Sanders MH, et al;
Sleep AHEAD Research Group of Look AHEAD
Research Group. A randomized study on the
effect of weight loss on obstructive sleep apnea
among obese patients with type 2 diabetes: the
Sleep AHEAD study. Arch Intern Med 2009;169:
16191626
25. Rubin RR, Peyrot M, Gaussoin SA, et al;
Look AHEAD Research Group. Four-year analysis of cardiovascular disease risk factors, depression symptoms, and antidepressant
medicine use in the Look AHEAD (Action for
Health in Diabetes) clinical trial of weight
loss in diabetes. Diabetes Care 2013;36:
10881094
26. Gregg E, Chen H, Wagenknecht L, et al.;
Look AHEAD Research Group. Association of
an intensive lifestyle intervention with remission of type 2 diabetes. JAMA 2012;308:2489
2496
27. Rejeski WJ, Ip EH, Bertoni AG, et al; Look
AHEAD Research Group. Lifestyle change and
mobility in obese adults with type 2 diabetes.
N Engl J Med 2012;366:12091217
28. Greenway FL, Ryan DH, Bray GA, Rood JC,
Tucker EW, Smith SR. Pharmaceutical cost savings of treating obesity with weight loss medications. Obes Res 1999;7:523531
29. Potteiger CE, Paragi PR, Inverso NA, et al.
Bariatric surgery: shedding the monetary
weight of prescription costs in the managed
care arena. Obes Surg 2004;14:725730
30. Davis WA, Bruce DG, Davis TME. Economic
impact of moderate weight loss in patients with
type 2 diabetes: the Fremantle Diabetes Study.
Diabet Med 2011;28:11311135
31. Redmon JB, Bertoni AG, Connelly S, et al;
Look AHEAD Research Group. Effect of
the look AHEAD study intervention on medication use and related cost to treat cardiovascular disease risk factors in individuals with
type 2 diabetes. Diabetes Care 2010;33:
11531158
32. Koster I, von Ferber L, Ihle P, Schubert I,
Hauner H. The cost burden of diabetes mellitus:
the evidence from Germanydthe CoDiM study.
Diabetologia 2006;49:14981504
33. Wolfenstetter SB. Future direct and indirect
costs of obesity and the inuence of gaining
weight: results from the MONICA/KORA cohort
studies, 1995-2005. Econ Hum Biol 2012;10:
127138
34. Little RJA. Pattern-mixture models for multivariate incomplete data. J Am Stat Assoc 1993;
87:12271237