eight Management Using the InternetRandomized Controlled Trial
hristine M. Hunter, PhD, Alan L. Peterson, PhD, Lisa M. Alvarez, PhD, Walker C. Poston, PhD, MPH,ntoinette R. Brundige, MS, C. Keith Haddock, PhD, David L. Van Brunt, PhD, John P. Foreyt, PhD
ackground: Most weight-loss research targets obese individuals who desire large weight reductions.However, evaluation of weight-gain prevention in overweight individuals is also critical asmost Americans become obese as a result of a gradual gain of 1–2 pounds per year overmany years.
ethod: This study evaluated the efficacy of an Internet-based program for weight-loss andweight-gain prevention with a two-group, prospective, randomized controlled trial. Amilitary medical research center with a population of 17,000 active-duty military personnelsupplied 446 overweight individuals (222 men; 224 women) with a mean age of 34 yearsand a mean BMI of 29. Recruitment and study participation occurred 2003–2005 and datawere analyzed in 2006. Participants were randomly assigned to receive the 6-monthbehavioral Internet treatment (BIT, n�227) or usual care (n�224). Change in bodyweight, BMI, percent body fat, and waist circumference; presented as group by timeinteractions, were measured.
esults: After 6 months, completers who received BIT lost 1.3 kg while those assigned to usual caregained 0.6 kg (F(df�366)�24.17; I�0.001). Results were similar for the intention-to-treatmodel. BIT participants also had significant changes in BMI (–0.5 vs �0.2 kg/m2;F(df�366)�24.58); percent body fat (–0.4 vs �0.6%; F(df�366)�10.45); and waist circumfer-ence (–2.1 vs –0.4 cm; F(df�366)�17.09); p�0.001 for all.
onclusions: Internet-based weight-management interventions result in small amounts of weight loss,prevent weight gain, and have potential for widespread dissemination as a populationhealth approach.
he average body weight of adults in the UnitedStates has increased dramatically over the past 3decades1–3 and nearly two thirds of adults are
ow overweight (BMI�25) or obese (BMI�30). Mosteight-management studies have targeted significantlyverweight or obese individuals who desire moderate to
arge reductions in weight.4–6 However, most Ameri-
rom the Division of Diabetes, Endocrinology, and Metabolic Dis-ases, National Institute of Diabetes and Digestive and Kidneyiseases (Hunter), Bethesda, Maryland; University of Texas Healthcience Center (Peterson); Department of Psychology, Wilford Halledical Center (Peterson, Alvarez, Brundige), San Antonio, Texas;niversity of Missouri, Kansas City School of Medicine (Poston,addock), Kansas City, Missouri; University of Tennessee Health
ciences Center (Van Brunt), Memphis, Tennessee; Eli Lilly andompany (Van Brunt), Indianapolis, Indiana; Baylor College ofedicine (Alvarez, Brundige, Foreyt), Houston, TexasAddress correspondence and reprint requests to: Christine M.unter, PhD, Division of Diabetes, Endocrinology, and Metaboliciseases, National Institute of Diabetes and Digestive and Kidney
m J Prev Med 2008;34(2)2008 American Journal of Preventive Medicine • Published by
ans become overweight or obese as a result of aradual weight gain of 1–2 pounds per year over manyears.7–9 Although clinic-based behavioral weight-lossrograms can produce safe and meaningful weight
oss,10,11 they are inefficient for the escalating needs ofhe U.S. population.12
The Internet offers a potential platform for a broaderpproach to weight management,13,14 and most adultsrefer the Internet over face-to-face professional encoun-
ers.15 Several randomized trials evaluating Interneteight-management interventions have demonstrated ef-cacy for producing meaningful weight loss and prevent-
ng weight regain.16–22 In sum, these researchers foundhat increased contact and tailored interaction althoughhe web, including e-mail, tailored messages, chat rooms,nd food/exercise diary feedback, improved weight-lossnd weight-maintenance outcomes compared to less-nteractive educational web-based formats.16,23 However,
ost of the existing research has been conducted incademic centers with samples that are over forty, female,
redominantly obese, and have a co-morbid health con-
1190749-3797/08/$–see front matterElsevier Inc. doi:10.1016/j.amepre.2007.09.026
ition such as diabetes.16–18 In the one published studyhat included a large and diverse sample, the findingsere limited by the poor retention at follow-up (20%) and
he fact that weight outcomes were based on participants’elf-reports.22
Additional research is thus required to examine theffectiveness of an Internet weight-management programn a more diverse sample and in people without co-
orbid health conditions. Evaluation in diverse samplesnd other settings will provide information about theeplicability and generalizability of the early findings.urther, research is needed to test Internet programs formall-to-moderate weight loss and prevention of weightain as this may be a more realistic population-health–ased goal.14 The current study evaluated the effective-ess of a behavioral treatment program for weightanagement using the Internet as compared to usual
are in a diverse sample of primarily overweight adults.t was hypothesized that those in the Internet-basedrogram would, as a group, demonstrate the preven-ion of weight gain or small-to-moderate weight losses.t was further hypothesized that participants in thesual-care condition would not show any weight lossnd would, in fact, gain weight over the 6-month periodrom baseline to reassessment.
ethods
articipants
he U.S. Air Force (USAF) was identified as a favorable popu-ation for study because of the high percentage of men, ethnic
inorities, and younger adults. As noted, these groups are oftennderrepresented in most Internet weight-management trials.espite an emphasis on physical fitness, 59.8% of USAF person-el are overweight and 13.4% of these are obese.24 Similar toivilian populations,7–8 USAF personnel gained an average of–2 pounds per year over the past 5 years,9 and the rate ofverweight has increased in young USAF recruits by 1% perear.25
The participants were 446 healthy adults (222 men, 224omen) with a mean BMI of 29 kg/m2 and a mean age of 34.articipants were recruited between June 2003 and October005 at Lackland and Randolph Air Force Bases and Brooksity Base. These bases are all in San Antonio TX, and at the
ime of the study, included a population of approximately7,000 active-duty military personnel. Recruitment messagesent out to active-duty members through a series of emaildvertisements and flyers soliciting individuals who wanted toose weight or prevent weight gain. Based on crude estimatesegarding the penetration of recruitment messages, it is esti-ated that the recruitment message was seen by at least half of
he available active-duty population (n�8500) of which approx-mately 5100 individuals (60%) were likely to be overweight. Asuch, the 446 consented participants represent an interventioneach of about 9% of the population that might have benefitedrom a weight-management intervention.
Participants were initially screened over the phone forligibility. Eligibility criteria included an age between 18 and
5 years, weight within 5 pounds or above their maximum i
20 American Journal of Preventive Medicine, Volume 34, Num
llowable weight (MAW) for the USAF (MAW�BMI�25 inomen, �27.5 in men for most heights),26 availability of aersonal computer with Internet access, and plans to remain
n the local area for 1 year. Participants were ineligible if theyad lost more than 10 pounds in the previous 3 months; usedrescription or over-the-counter weight-loss medications inhe previous 6 months; had any physical activity restrictions;ad a history of myocardial infarction, stroke, or cancer in the
ast 5 years; reported diabetes, angina, or thyroid difficulties;r had orthopedic or joint problems that would prohibitxercise. Women were excluded if they were currently preg-ant or breast-feeding, or had plans to become pregnant in
he next year.
esign
he study used a two-group parallel randomized controlledrial design. Eligible individuals were scheduled for baselinessessment where they completed informed consent, ques-ionnaires, and were measured for height, weight, waistircumference, and body fat. Participants were then random-zed to usual care or behavioral Internet therapy (BIT) plussual care. All measures were repeated at a 6-month post-aseline randomization. Due to the nature of the interven-ion, study staff and participants could not be blinded toroup assignment. Prior to study initiation, all interventionnd study procedures were reviewed and approved by thenstitutional Review Boards of Baylor College of Medicinend the U.S. Air Force and Army Surgeons General.
rocedures for Usual Care
t an USAF base, usual care is a more conservative compari-on condition than in most settings. USAF members see theirrimary care provider at least once annually for a preventiveealth visit that includes an assessment of diet and weight.ach base included at least one fitness center, weight loss andealthy cooking classes, available nutrition consultants, andpportunities for individual fitness assessments and recom-endations. Further, USAF members are expected to work
ut with their unit a minimum of three times per week andre tested annually for fitness (timed run, push-ups, andit–ups), BMI, and waist-circumference standards. Finally,here are occupational pressures that may enhance motiva-ion to use the available resources, as failure to meet BMI,aist, or fitness standards can result in loss of a promotion orcareer in the USAF.
rocedures for Behavioral Internet Therapy (BIT)
n addition to usual care, the BIT participants attended ann-person orientation where they were given instructionsbout the components of BIT to include details about calcu-ating calories and energy expenditure, how to submit elec-ronic food and exercise diaries, when to expect weeklyounselor feedback on their Internet diaries, and establish-ent of any weight-loss and calorie goals. Although a primary
im of the study was prevention of weight gain, most partici-ants also were interested in losing weight. Early in recruit-ent it was found that prevention of weight gain was rarely a
ompelling reason to join the study, but when combined withesources that could also help with weight loss, potentialarticipants were significantly more interested in participat-
ng. The BIT program included behavioral, dietary, and
ber 2 www.ajpm-online.net
eswg1te
mwtwwassmpaeaducetpdwqtItwsEPbwcI
t8tbttatbipptawr
O
Tp
cnmomBsiuaQ
S
RrasSptriattwtidera
D
SMlsccmstawttt
RP
Bpruflc
F
xercise recommendations designed to facilitate the loss ofmall-to-moderate amounts of weight and prevent futureeight gain. Based on starting weight, participants wereenerally asked initially to restrict their calorie intake to200–1500 kcals per day, keeping fat intake below 30% of theotal, and to increase physical activity until their estimatedxpenditure was at least 1000 kcals a week.Participants were instructed to log-in and complete self-onitoring food and exercise diaries at least five times aeek. They were instructed to monitor their weight weekly
hrough a weight-tracking chart. BIT participants receivedeekly personalized feedback on the food, exercise, andeight information they submitted online. Participants weressigned weekly lessons on the website that included commontrategies associated with behavioral weight loss such astimulus control, behavior modification, and stress manage-ent.27,28 This approach was taken because of the overlap-
ing goals between the study and the participants (weight lossnd/or weight gain prevention) and because there is consid-rable overlap in strategies used for weight-gain preventionnd weight loss. In regard to dietary and exercise recommen-ations for prevention and weight loss, the differences aresually related to the magnitude of the recommendedhange (e.g., small versus large changes in caloric intake orxercise). Each lesson included interactive quizzes, and accesso subsequent lessons was dependent on completion of therevious lesson in an attempt to keep the website contentynamic. The lessons took approximately 20–30 minutes aeek for the participants to complete. The only other re-uired time on the website was the time needed to completehe self-monitoring diaries. Each participant was assigned onenternet counselor for the duration of the intervention andhese counselors spent 10–15 minutes per participant eacheek providing written feedback in response to submitted
elf-monitoring diaries. Participants were given The Lifestyle,xercise, Attitudes, Relationships, and Nutrition (LEARN)rogram for Weight Management 2000,29 an evidence-basedehavioral modification approach to weight management,ith about 5 minutes of weekly assigned readings designed toomplement the 24 weeks of progressive and interactiventernet lessons.
Participants were also scheduled for two brief motiva-ional interviewing30,31 telephone calls scheduled at 4- and-weeks post-baseline. Motivational interviewing is a direc-ive and patient-centered counseling style for elicitingehavior change.31–36 It was selected as the counseling style for
he phone calls because Internet interventions have been showno perform better when combined with individualized supportnd feedback,16,23 and recent meta-analyses found that motiva-ional interviewing is effective for changing diets and exerciseehaviors and reducing BMI.32,35 The implicit goal is to
ncrease the client’s awareness of the discrepancy betweenresent behaviors and future goals. The same counselor whorovided feedback on the self-monitoring diaries was assignedo make the phone calls. Each call lasted about 15 minutes,nd the content was driven by the individual’s needs (e.g.,ork on decision balance, support/encouragement, collabo-ative problem solving).
utcome Measures
he primary outcomes were change in body weight (kg and
ercent change from baseline), percent body fat, and waist m
ebruary 2008
ircumference. These measures were taken by study person-el at baseline and 6 months. Weight and height wereeasured on a calibrated scale with participants in uniform
r light street clothes and without shoes. Body fat waseasured through bioelectrical impedance with the Tanitaody Composition Analyzer.37 Waist circumference was mea-
ured with a Gulick tape at the umbilicus.38 Other measuresncluded tracking log-ins to the website, eating behaviorssing the Brief Fat, Fruit, Vegetable, and Fiber Screeners,39
nd physical activity using the International Physical Activityuestionnaire (IPAQ).40,41
ample Size and Randomization
andomization to BIT or usual care was conducted using aandom-numbers table and was stratified based on gendernd whether participants were above or below their MAW aspecified by USAF guidelines in Air Force Instruction 40-502.ample size was estimated using data from one of the firstublished Internet-based intervention studies for obesity.16 In
hat study, by Tate and colleagues, intention-to-treat (ITT)esults demonstrated that patients receiving Internet behav-or therapy lost 2.9�4.4 kg compared to those randomized ton Internet education group, who lost 1.3�3.0 kg. Givenhese data, the larger variance estimate was used conserva-ively. Also, a conservative estimate of attrition, of up to 40%,as used given the low-intensity nature of the BIT interven-
ion. Power calculations (SPSS Sample Power, Release 2.0)ndicated that there would be 85% power to detect a 1.6-kgifference with a minimum of 270 total participants at thend of intervention period. Thus, the study recruited andandomized 446 participants to accommodate up to 40%ttrition or, if attrition was less, a smaller effect size.
ata Analysis
tatistical analyses were performed using SPSS, version 14.0.eans�standard deviation scores or percentages were calcu-
ated for all baseline demographic variables. Repeated mea-ures ANOVA (RANOVA) was used to examine 6-monthhanges in weight, BMI, body fat percentage, and waistircumference using both completers and ITT as recom-ended by CONSORT Guidelines42 using the baseline ob-
ervation carried forward (BOCF).43 RANOVA also was usedo examine changes in diet and physical activity. Thesenalyses were conducted only on completers because theyere secondary outcomes. All RANOVA models are two-
reatment group (BIT, usual care) � 2-time (pretest, post-est) models and the group � time interaction test was usedo determine treatment effects.
esultsarticipants
aseline participant demographic characteristics areresented in Table 1. A total of 227 participants wereandomized to receive the BIT intervention and 224 tosual care. However, five participants were excludedrom analyses due to pregnancy after randomization,eaving 224 and 222 participants in the BIT and usual-are groups, respectively. Both groups included 50%
en and women. Participants had similar distribution
Am J Prev Med 2008;34(2) 121
iw(atard(tfi
6aawd
vhp
I
StstlfrsPoai
T
C
nAGEMPYPEa
d ity.B .
T
O
WBWB5
P
M
F
a
b
B
1
n age and rank as the eligible population, but moreomen and more minority participants were recruited50% of the sample were women, although womenccount for only 25% of active-duty USAF and 42% ofhe sample were minorities, while minorities representbout 34% of the USAF active-duty members in thategion). There were no significant differences in theistribution of demographic factors or clinical variablessee Tables 1 and 2) between participants in thereatment or usual-care groups (p�0.05). See Figure 1or CONSORT screening, recruitment, and completernformation.
There were no significant differences in follow-up atmonths across groups (see Table 2), with an overall
ttrition rate of 21.0% for the Internet-treated groupnd 14.0% for those in usual care (�2�3.81; p�0.051)ith a combined dropout rate of 17.1%. Those whoropped out were somewhat younger (32.4�7.8 years
able 1. Baseline characteristics (M�SD or percentages) by
haracteristic*
ge (years)ender (% female)thnicity (% Caucasian)a
arital status (% married or partnered)ercent enlisted (%)ears of service (years)lan to retire from AF (% yes)ducation (% high school or some college)
Participants classified their own race based on U.S. Census cateemonstrated variable response to weight loss based on race/ethnicIT, behavioral Internet therapy; M, mean; SD, significant deviation
able 2. Pre-test, post-test values and changes in primary ou
utcomes
BIT
Pre-test Post-test
eight (kg) 87.4�15.6 85.5�15.8MI (kg/m2) 29.4�3.0 28.8�3.3aist circumference (cm) 94.5�11.0 92.2�11.6ody fat percentage 34.5�6.8 33.9�7.3% or more weight loss
(% yes)— —
ercent gained weight — —No follow-up at 6
months (% dropout)— —
Block dietary screenerb
eat and snacks Screenerscore
23.6�7.9 19.0�8.1
Percent of caloriesfrom fat
35.1�4.8 32.3�4.8
ruit–vegetable–beansScreener score
13.7�5.5 15.8�5.9
Dietary fiber score 15.8�4.5 17.4�5.0IPAQ (Total
MET-minutes/week)2787.7�2863.0 2765.0�3069.4
Post-test data presented for completers only. Bolded differences we
RANOVA models only run for main dietary screener scores.IT, behavioral Internet therapy; IPAQ, International Physical Activity Qu
22 American Journal of Preventive Medicine, Volume 34, Num
s 34.3�7.2 years; F�4.37, p�0.037) and slightlyeavier (30.1�3.4 kg/m2 vs 29.2�2.8 kg/m2; F�5.78,�0.017) at baseline.
ntervention Fidelity Check
everal process variables were examined first to ensurehat the intervention was delivered to participants as-igned to BIT. The frequency of receiving the motiva-ional phone calls during the treatment period was calcu-ated; 209 (93.4%) of treated participants were availableor the 4-week motivational phone call, and 176 (78.4%)eceived the 8-week call. Next, website use by participantelf-report and total actual website logins was examined.articipants reported their weekly website use as less thannce (42.4%), 1–2 (22.6%), 3–4 (18.1%), 5–7 (9.6%),nd 7� (7.3%) times per week. BIT participants loggednto the website an average of 49.1 times over the treat-
ent period, with a range of 1–707 logins. Finally, actualogins, as recorded by the website, were converted intouartiles of use with the following distribution of total
ogins: Quartile 1�3.8�2.5 logins; Quartile 2�15.6�4.0ogins; Quartile 3�35.4�8.8 logins; and Quartile�133.6�83.7 logins.The relationship between weight loss and self-reported
ebsite use, actual logins, and the frequency of diaryse among those assigned to receive the BIT interven-ion was examined. Table 3 presents the correlationsnd significance levels. There was a statistically signifi-ant and modest relationship between weight loss andelf-reported website use, actual website login fre-uency, and diary use, indicating that greater use of theebsite by either metric was associated with significantlyore weight loss over the 6-month treatment period. In
ddition, Figure 2 presents weight losses (kg) by cate-ory of self-reported use and by actual login quartilemong BIT participants. Although self-report and ob-
igure 1. Flow diagram.IT, behavioral Internet treatment; ITT, intention to treat;C, usual care.
able 3. Intervention fidelity check among treated participa
Number of loginsfrom site counter
Self-reportedlogins
elf-reported logins 0.562 (�0.001) —umber of logins fromsite counter
Pearson correlations between actual and self reported log-ins and change
ebruary 2008
ective data often differ by a wide margin, the similarityn self-report and automated usage in this study mayave been partly due to the fact that this military cohortas aware that their logins were tracked, which mayave led to more accurate, or at least more careful,elf-assessment.
hanges in Body Composition
articipants assigned to the BIT intervention lost.3�4.1 kg while those assigned to usual care gained
igure 2. Weight change by category of self-reported websitese and site logins.OCF, baseline observation carried forward.
nts (p
001)
001)001)001)
001)
in body composition from baseline to 6-months.
Am J Prev Med 2008;34(2) 123
0(uBcpp((esa
fau(tlfi
m5uptwiwwgst
C
CS
ww(BmuIast(
P
AttfwaiImtmhWcopwtlwbwihi
D
TiiWatsowBipp
Ts
O
B
G
R
a
sBHawB
1
.6�3.4 kg; this difference was statistically significantF(df�1,366)�24.17; p�0.001). Results were similar whensing a BOCF model, with participants who received theIT intervention losing 1.0�3.7 kg while those in usualare gained 0.5�3.1kg (F(df�1,444)�23.12; p�0.001). BITarticipants also experienced more favorable body com-osition changes on all measures (p�0.001), i.e., BMI�0.5 vs �0.2 kg/m2; F(df�1,366)�24.58), percent body fat�0.4 vs �0.6%; F(df�1,366)�10.45), and waist circumfer-nce (�2.1 vs �0.4 cm; F(df�1,366)�17.09). Table 2 pre-ents changes in body composition by treatmentssignment.
Table 4 presents weight-loss change scores (M�SD)or study completers stratified by BMI level, gender,nd race. Weight-loss differences between BIT andsual care were significant in all stratified analysesp�0.01). However, there were no significant interac-ions between treatment group assignment and BMIevel, race, or gender, indicating that none of theseactors significantly moderated the effects of the BITntervention.
Significantly more BIT participants met a 5% orore weight-loss criterion, with 22.6% losing at least
% of initial body weight compared to only 6.8% ofsual-care participants (�2�18.59; p�0.001). Becauserevention of weight gain was a key focus of this study,he participants also were classified according tohether they maintained or lost weight during the
ntervention period compared to those who gained anyeight. Participants receiving the BIT interventionere significantly less likely to be classified as weightainers (41.8%) at the post-intervention 6-month as-essment as compared to those in the usual-care condi-ion (59.7%; �2�11.75; p�0.001).
hanges in Diet and Physical Activity
hanges in the primary scales of the Block Rapid Food
able 4. Primary 6-month outcome change scores (M�SD)tratified by BMI level, gender, and racea
Differences in weight by treatment status were significant in analysestratified by BMI, gender, and race. Weight loss differences betweenIT versus usual care were significant in all stratified analyses.owever, there were no significant interactions between treatment
ssignment and BMI level, gender or race, but power to asses theseas low (i.e., �0.50).IT, behavioral Internet treatment.
creener to assess fat and fruit and vegetable intake39 c
24 American Journal of Preventive Medicine, Volume 34, Num
ere examined. Significant group by time interactionsere found for both the Meat and Snacks Screener ScoreF(df�1,366)�4.034; p�0.045) and the Fruit–Vegetable–eans Screener Score (F(df�1,366)�9.921; p�0.002) at 6onths. Mean changes scores for both the BIT and
sual-care groups are provided in Table 2. Finally, thePAQ was used to evaluate changes in level of physicalctivity. In total MET-minutes/week, there were noignificant differences between BIT and usual-care par-icipants in physical activity change at 6 monthsF(df�1,307)�0.036; p�0.850).
articipant Ratings of Treatment Components
t the 6-month assessment, BIT participants were askedo complete a questionnaire about the helpfulness ofhe intervention components and to provide generaleedback. Helpfulness was rated on a 10-point scaleith 0 indicating that they found the component “nott all helpful,” 4–5 “somewhat helpful,” and 9 indicat-ng that they found the component “very helpful.” Thenternet-based food and exercise diaries were rated asost helpful (6.6�2.6 and 6.3�2.3, respectively), while
he Internet weight diaries, weekly lessons, and LEARNanual were rated as slightly better than somewhat
elpful (5.9�2.6, 5.6�2.3, and 5.5�2.5, respectively).ritten feedback from participants indicated that the
ounselor feedback on diaries was the most useful partf the website but, as in most weight-managementrograms, there was also feedback that self-monitoringas very time-consuming. Some individuals indicated
hat they chose a web-based program because they wereooking for an easier way to manage their weight andere disappointed that it still required considerableehavioral change. The feedback about the phone callsas almost universally positive with participants indicat-
ng that the personal interaction and encouragementelped to keep them motivated and engaged in the
ntervention.
iscussion
he Internet has become a primary source for healthnformation, and is a promising tool for the delivery ofnterventions to change health behavior.13,14,16–18,20–23
hile the weight change in this study was not as larges in early Internet-based weight-management studies,hat may be due to the considerable differences in theamples studied and the added emphasis in this studyn prevention of weight gain in a predominately over-eight population. However, despite these differences,IT was effective in achieving weight loss and prevent-
ng weight gain. As compared to usual care, the BITarticipants had significantly better outcomes on allrimary measures including weight loss, BMI, waist
ircumference, body fat, and prevention of weight gain.
ber 2 www.ajpm-online.net
Aawtwbstoww
stTcdipadiwaFdunihaio
aotopwmramtburtiwtpatgh
uum
hmodksfTpiaaAiamtcfibbfcBct
ogmneastot
ittdliglcbcsph
F
lthough most BIT participants did not lose largemounts of weight, there were medically significanteight losses (�5%) in three times as many of the
reated population, and there were fewer who gainedeight as compared to the usual-care group. As mighte expected, more frequent use of the website andelf-monitoring led to better outcomes. On average,hose participants who used the website as instructed (5r more days a week) lost 4.17–5.69 kg. at 24 weekshile those who logged in less than once a week gainedeight.This study included several strengths. The sample
ize was more than double previous Internet weight-lossrials that directly measured body composition.16–21
he sample was also younger, included an equal per-entage of men and women, and had more ethniciversity than in previous Internet studies, which is
mportant when examining an intervention’s potentialublic health benefit, as increasing weight is a concerncross genders and ethnicities. The lack of significantifference in outcome by race or ethnicity is interesting
n light of the differences usually seen in traditionaleight-loss trials. This trial also had good retention aslmost 83% of the sample was evaluated at 6 months.inally, the sample may be a better representation of aisseminable weight-management approach because,nlike many trials, no run-in phase was instituted ando reimbursement incentives were provided for partic-
pation. Although a run-in phase and incentives areighly valuable for assuring retention and enhancingdherence, these strategies were not used in this studyn order to better measure effectiveness in the contextf a population health model.The primary limitation of the study was that the
verage participant in the active intervention achievednly a modest 1.3-kg weight loss. In part this may be dueo lower average starting weight as compared to mostther Internet weight studies and the fact that partici-ants were recruited to prevent weight gain and/or loseeight. Another limitation was the lack of objectiveeasurement of physical activity. Although the self-
eport measure used has reasonable validity and reli-bility, a more objective measure might have revealed aoderating influence of physical activity.40,41 Finally,
he inclusion of only active-duty USAF members coulde considered an additional limitation. Despite theniqueness of the sample, the authors believe theseesults may have important implications for generaliza-ion to the U.S. population due the much greaternclusion of men, minorities, younger adults, and over-eight (versus primarily obese) participants. However,
he advantage of the diversity must be balanced by theotential limitations of a military sample. For example,ll participants were employed, had a shared occupa-ional mission, had free access to health care, hadraduated from high school, and were pre-screened for
ealth and weight prior to entry into the USAF. Also, o
ebruary 2008
nlike most of the U.S population, USAF members arender occupational pressure to be physically fit andanage their weight.Despite some of the limitations, the Internet clearly
as promise as a platform for population-based weightanagement and the prevention of weight gain. Previ-
us research has consistently indicated that regularocumentation of weight-related behaviors, such aseeping a daily food diary,45 are strong predictors ofuccess in weight management and this study offersurther support for the importance of self-monitoring.he Internet also allows for immediate feedback onrogress, which can be a powerful motivator for chang-
ng or maintaining health-related behaviors. However,s in all weight-management programs, success is usu-lly dependent on the use of the provided strategies.dvertising the Internet as an “easy” way to participate
n a weight-management program may underplay thectual work required by participants. In the future, itay be important to emphasize that the ease is related
o flexibility of delivery and not necessarily the behaviorhange. Finally, there has been growing interest innding ways to further automate Internet-based feed-ack as a way to enhance the potential for public healthenefit and control personnel costs. The participanteedback suggests that automation must be approachedarefully and cautiously as the personalized aspects ofIT (feedback on self-monitoring diaries and phonealls) had a positive relationship to outcome and par-icipant satisfaction.
Additional research is needed to determine what typesf patients may benefit most from Internet-based pro-rams and to examine factors necessary to improve oraintain weight management. For example, research is
eeded to examine potential moderators of treatmentffectiveness such as computer experience, health liter-cy, and type of occupation. Also, cost-effectiveness re-earch is needed that compares Internet-based interven-ion to traditional in-person intervention or other formsf minimal participant contact such as compact disks andelephone only.
The Internet allows for widespread delivery and dissem-nation of evidence-based weight-management programshat require less personnel time as compared to tradi-ional weight-management programs. This type of re-uced face-to-face contact intervention may be particu-
arly beneficial for weight-gain prevention, which isncreasingly recognized as an important intervention tar-et given the general difficulty in maintaining weightoss.12,46–48 The amount of weight gained in the usual-are group (0.5 kg.) was similar to what has previouslyeen reported in large longitudinal studies of bothivilian and military populations.8,9 Although most re-earch targets weight loss, weight-gain prevention ap-roaches, such as those used in the present study, mayelp alter the epidemic increases in overweight-and
besity in the U.S. population.44
Am J Prev Med 2008;34(2) 125
TUas
tNF
t
R
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
1
he present study was funded through a grant funded by the.S. Department of Defense, U.S. Army Medical Research
nd Material Command, and the Peer Review Medical Re-earch Program (DAMD17-02-1-0180).
The opinions and assertions herein are the private views ofhe authors and are not the official policy or position of theational Institutes of Health or the Department of the Airorce.No financial disclosures were reported by the authors of
The continuing epidemic of obesity in the United States. JAMA2000;284:1650–1.
2. Flegal KM, Graubard BI, Williamson DF, Gail MH. Excess deaths associatedwith underweight, overweight, and obesity. JAMA 2005;293:1861–7.
3. Ogden C, Carroll M, Curtin L, McDowell M, Tabak C, Flegal K. Prevalenceof overweight and obesity in the United States, 1999–2004. JAMA 2006;295:1549–55.
4. Curioni CC, Lourenco PM. Long-term weight loss after diet and exercise: asystematic review. Int J Obes 2005;29:1168–74.
5. Li Z, Maglione M, Tu W, et al. Meta-analysis: pharmacologic treatment ofobesity. Ann Intern Med 2005;142:532–46.
6. Shaw K, O’Rourke P, Del Mar C, Kenardy J. Psychological interventions foroverweight or obesity. Cochrane Database Syst Rev 2005;(2):CD003818.
7. Brown WJ, Williams L, Ford JH, Ball K, Dobson AJ. Identifying the energygap magnitude and determinants of 5-year weight gain in midage women.Obes Res 2005;13:1431–41.
8. Lewis CE, Jacobs DR Jr, McCreath H, et al. Weight gain continues in the1990s: 10-year trends in weight and overweight from the CARDIA study.Coronary Artery Risk Development in Young Adults. Am J Epidemiol2000;151:1172–81.
9. Robbins AS, Chao SY, Baumgartner N, Runyan CN, Oordt MS, Fonseca VP.A low-intensity intervention to prevent annual weight gain in active duty AirForce members. Mil Med 2006;171:556–61.
0. Wing RR. Behavioral approaches to the treatment of obesity. In: Bray G,Bouchard C, James P, eds. Handbook of obesity. New York: Marcel DekkerInc, 1998:855–73.
1. Wing RR. Behavioral weight control. In: Wadden TA, Stunkard AJ, eds.Handbook of obesity treatment. New York: The Guilford Press; 2002:301–16.
2. Gill T, King, L, Caterson I. Obesity prevention: necessary and possible. Astructured approach for effective planning. Proc Nutr Soc 2005;64:255–61.
3. U.S. Department of Health and Human Services (USDHHS), Office ofDisease Prevention and Health Promotion. Expanding the reach andimpact of consumer e-health tools. Available online at: http://www.health.gov/communication/ehealth/ehealthTools/default.htm.
4. Winett RA, Tate DF, Anderson ES, Wojcik JR, Winett SG. Long-term weightgain prevention: A theoretically based Internet approach. Prev Med2005;41:629–41.
5. Sherwood NE, Morton N, Jeffrey RW, French SA, Neumark-Sztainer D,Falkner NH. Consumer preferences in format and type of community-based weight control. Am J Health Promot 1998;13:12–8.
6. Tate DF, Wing RR, Winett RA. Using Internet technology to deliver abehavioral weight loss program. JAMA 2001;285:1172–7.
7. Tate DF, Jackovny EH, Wing RR. Effects of Internet behavioral counselingon weight loss in adults at risk for type 2 diabetes: A randomized trial. JAMA2003;289:1833–6.
8. Tate DF, Jackovny EH, Wing RR. A randomized trial comparing humane-mail counseling, computer-automated tailored counseling and no coun-seling in an Internet weight loss program. Arch Intern Med 2006;66:1620–5.
9. Harvey-Berino J, Pintauro S, Buzzell P, Gold EC. Effect of Internet support
on the long-term maintenance of weight loss. Obes Res 2004;12:320–9.
26 American Journal of Preventive Medicine, Volume 34, Num
0. Williamson DA, Martin PD, White MA, et al. Efficacy of an Internet-basedbehavioral weight loss program for overweight adolescent African-Americangirls. Eat Weight Disord 2005;10:193–203.
1. Williamson DA, Walden HM, White MA, et al. Two-year Internet-basedrandomized controlled trial for weight loss in African-American girls.Obesity 2006;14:1231–43.
2. Rothert K, Strecher VJ, Doyle LA, et al., Web-based weight managementprograms in an integrated health care setting: A randomized, controlledtrial. Obesity 2006;14:266–72.
3. Tate DF, Zabinski MF. Computer and Internet applications for psycholog-ical treatment: update for clinicians. J Clin Psychol 2004;60:209–20.
4. Bray RM, Hourani LL, Rae KL, et al. 2005 department of defense survey ofhealth related behaviors among military personnel. 2006. Available onlineat: www.tricare.osd.mil.
5. Poston WC, Haddock CK, Peterson AL, et al. Comparison of weight statusamong two cohorts of U.S. Air Force recruits. Prev Med 2005;40:602–9.
6. Haddock CK, Poston WC, Klesges RC, Talcott GW, Lando H, Dill P. Anexamination of body weight standards and the association between weightand health behaviors in the United States Air Force. Mil Med 1999;164:51–4.
7. Foreyt JP, Goodrick GK. Attributes of successful approaches to weight lossand control. Appl Prev Psychol 1994;3:209–15.
8. Perri MG, Fuller PR. Success and failure in the treatment of obesity: Wheredo we go from here? Exerc Nutr Health 1995;4:255–72.
9. Brownell KD. The LEARN program for weight management 2000. DallasTX: American Health Publishing Company, 2000.
0. Miller WR, Rollnick S. Motivational interviewing: preparing people tochange addictive behavior. 2d edn. New York: Guilford Press, 2003.
1. Draycott S, Dabbs A. Cognitive dissonance. 2: A theoretical grounding ofmotivational interviewing. Br J Clin Psychol 1998;37:355–64.
2. Burke BL, Arkowitz H, Menchola M. The efficacy of motivational interview-ing: a meta-analysis of controlled clinical trials. J Consult Clin Psychol2003;71:843–61.
4. Miller WR, Rollnick S. Motivational interviewing: preparing people tochange addictive behavior. New York: Guilford Press, 1991.
5. Ruback S, Sandbaek A, Lauritzen T, Christensen B. Motivational Interview-ing: as systemic review and meta-analysis. Br J Gen Pract 2005;55:305–12.
6. Smith DE, Heckemeyer CM, Kratt PP, Mason DA. Motivational interviewingto improve adherence to a behavioral weight control program for olderobese women with MIDDM. A pilot study. Diabetes Care 1997;20:52–4.
7. Ritchie JD, Miller CK, Smiciklas-Wright H. Tanita foot-to-foot bioelectricalimpedance analysis system validated in older adults. J Am Diet Assoc2005;105:1617–9.
8. Lohman TG, Roche AF, Martorell R. Anthropometric standardizationreference manual. Champaign IL: Human Kinetics, 1988.
9. Block GG, Rosenbaum C, Jenson C. A rapid food screener to assess fat andfruit and vegetable intake. Am J Prev Med 2000;18:284–8.
0. Craig CL, Marshall AL, Sjostrom M, et al. International physical activityquestionnaire: 12-country reliability and validity. Med Sci Sports Exerc2003;25:1381–95.
1. Tehard B, Saris WH, Astrup A, et al. Comparison of two physical activityquestionnaires in obese subjects: The NUGENOB study. Med Sci SportsExerc 2005;37:1535–41.
2. Moher D, Schulz KF, Altman D, CONSORT Group. The CONSORTstatement: revised recommendations for improving the quality of reports ofparallel-group randomized trials. JAMA 2001;285:1987–91.
3. Ware JH. Interpreting incomplete data in studies of diet and weight loss.N Engl J Med 2003;348:2136–7.
4. Winker M. Measuring race and ethnicity: why and how? JAMA 2004;292:1612–4.
5. Burke LE, Warziski M, Starrett T, et al. Self-monitoring dietary intake:current and future practices. J Ren Nutr 2005;15:281–90.
6. Wing RR, Tate DF, Gorin AA, Raynor HA, Fava JL. A self-regulationprogram for maintenance of weight loss. N Engl J Med 2006;355:1563–71.
7. Grodstein F, Levine R, Troy L, Spencer T, Colditz GA, Stampfer MJ.Three-year follow-up in a commercial weight loss program. Arch InternMed 1996;156:1303–6.
