http://jama.ama-assn.org/cgi/content/full/285/9/1172
Permissions Reprints/E-prints
permissions@ama-assn.org reprints@ama-assn.org
http://pubs.ama-assn.org/misc/permissions.dtl
D
EVELOPING EFFECTIVE WEIGHT weight loss education Web site.
loss programs that are widely Design Randomized, controlled trial conducted from April to December 1999.
accessible is a health care pri- Setting and Participants Ninety-one healthy, overweight adult hospital employ-
ority given that more than ees aged 18 to 60 years with a body mass index of 25 to 36 kg/m2. Analyses were
54% of US adults are overweight or performed for the 65 who had complete follow-up data.
obese1 and that weight loss is recom- Interventions Participants were randomly assigned to a 6-month weight loss pro-
mended to reduce the health impact of gram of either Internet education (education; n = 32 with complete data) or Internet
obesity. Although group behavioral pro- behavior therapy (behavior therapy; n = 33 with complete data). All participants were
grams involving weekly clinic visits are given 1 face-to-face group weight loss session and access to a Web site with orga-
the most effective treatments available nized links to Internet weight loss resources. Participants in the behavior therapy group
for obesity, most adults would prefer to received additional behavioral procedures, including a sequence of 24 weekly behav-
lose weight without having to partici- ioral lessons via e-mail, weekly online submission of self-monitoring diaries with indi-
pate in a structured face-to-face treat- vidualized therapist feedback via e-mail, and an online bulletin board.
ment program.2 To accommodate the Main Outcome Measures Body weight and waist circumference, measured at 0,
needs of these individuals and to make 3, and 6 months, compared the 2 intervention groups.
obesity treatment more accessible, in- Results Repeated-measures analyses showed that the behavior therapy group lost
vestigators have explored alternative more weight than the education group (P = .005). The behavior therapy group lost a
methods for delivering weight loss pro- mean (SD) of 4.0 (2.8) kg by 3 months and 4.1 (4.5) kg by 6 months. Weight loss in
grams including mail-based correspon- the education group was 1.7 (2.7) kg at 3 months and 1.6 (3.3) kg by 6 months. More
dence programs3-5 interventions deliv- participants in the behavior therapy than education group achieved the 5% weight
loss goal (45% vs 22%; P = .05) by 6 months. Changes in waist circumference were
ered via telephone6 and television.7,8
also greater in the behavior therapy group than in the education group at both 3 months
Although these types of programs have (P=.001) and 6 months (P= .005).
typically produced smaller weight losses
Conclusions Participants who were given a structured behavioral treatment pro-
than standard group behavioral pro-
gram with weekly contact and individualized feedback had better weight loss com-
grams, they offer an important alterna- pared with those given links to educational Web sites. Thus, the Internet and e-mail
tive to face-to-face treatment. appear to be viable methods for delivery of structured behavioral weight loss programs.
Inthepastdecade,computer-mediated JAMA. 2001;285:1172-1177 www.jama.com
interventions have been developed for a
variety of behavior changes, including di-
etary change,9,10 smoking cessation,11 and the use of this technology as stand-alone has surged from 9% to 56% of adults
exercise12;however,fewstudieshavebeen obesity treatment or as an adjunct to stan- in the past 4 years.15 Conceptually, the
conducted using computers for the treat- dard therapy. Computer based programs Internet has distinct advantages for pro-
ment of obesity. Initial research in this can easily be adapted for use via the In-
areafocusedonusinghand-heldcomput- ternet, which has renewed interest in us-
Author Affiliations: Brown Medical School/Miriam
ers for entry of self-monitoring data, to ing this technology for weight loss. Hospital, Providence, RI (Drs Tate and Wing) and Vir-
giveautomatedfeedbackaboutcaloricval- Rapid increases in access to the In- ginia Polytechnic Institute and State University, Blacks-
burg, Va (Dr Winett).
ues and either provide praise for or in- ternet and the World Wide Web has Corresponding Author and Reprints: Deborah F. Tate,
structions on modifying eating contin- made it a viable and logical mode for PhD, Brown Medical School/Miriam Hospital, Weight
Control and Diabetes Research Center, 14 Third St,
gent on performance.13,14 Furthermore, public health intervention. The num- RISE Bldg, Providence, RI 02906 (e-mail: dtate
research has not expanded or developed ber of US adults who use the Internet @lifespan.org).
1172 JAMA, March 7, 2001—Vol 285, No. 9 (Reprinted) ©2001 American Medical Association. All rights reserved.
tion code and weight data were An e-mail message was sent to behav- Web site was tracked using a unique lo-
transmitted and stored using this code ior therapy participants each week dur- gin identification code for each partici-
rather than participant names; how- ing the 24-week program including a be- pant to record each login. An index of
ever, they were advised that the poten- havioral weight loss lesson and feedback. participants’ previous experience with the
tial existed for data and e-mail mes- The behavioral lesson included struc- Internet or e-mail was created by sum-
sages to be intercepted and read. tured guidance about a variety of weight ming the number of months partici-
The study Web site was accessible on loss topics on nutrition, exercise, or be- pants had used e-mail plus the number
the organization’s Intranet and pro- havioral self-regulatory strategies. In ad- of months they had used the Internet.
vided a brief review of basic informa- dition, each weekly e-mail included in-
tion related to weight loss and an or- dividualized feedback sent personally Statistical Analysis
ganized directory of selected Internet from the doctoral-level therapist. The Using an a level of .05 and power of
resources about diet, exercise, self- feedback included recommendations 80%, a sample size of 37 for each group
monitoring, and other resources that in- and reinforcement based on progress was needed to detect a 2.27-kg differ-
cluded behavioral topics including so- noted in the self-monitoring diary and ence between groups. Assuming an av-
cial support, stimulus control, and specifically addressed weight loss erage attrition rate of 20%,21 a sample of
managing stress. During the introduc- progress, dietary intake, and energy ex- at least 90 subjects was selected. To de-
tory group session, participants re- penditure. Recommendations and strat- tect changes in the outcomes of weight,
ceived a 1 hour lesson on behavioral egies for improvement also were pro- waist circumference, calorie intake, and
weight control. At this session, a stan- vided. In addition, the therapist expenditure, repeated-measures analy-
dard calorie restriction diet of 1200 to answered any questions raised by par- sis of variance (ANOVA) models were
1500 kcals per day and daily fat intake ticipants and provided general support used. All analyses were performed us-
of less than 20% of calories consumed and encouragement via the e-mail mes- ing the Statistical Package for the So-
was recommended. Participants were sage. Participants who did not send in cial Sciences (SPSS for Windows ver-
also instructed to increase gradually a log were sent a personal e-mail inquir- sion 10.05 SPSS Inc, Chicago, Ill).
their physical activity to burn a mini- ing about their progress and were en-
mum of 1000 kcals per week. The im- couraged to monitor and continue with RESULTS
portance of self-monitoring was stressed the program. Participants in the behav- A preliminary analysis showed that there
and participants in both groups were ior therapy group also had access to an were no differences between groups for
encouraged to use the self-monitoring electronic bulletin board to facilitate so- baseline measures of age, weight, BMI,
Web resources to keep track of their diet cial support among participants as- waist circumference, or Internet expe-
and exercise daily. However, only the signed to this intervention. rience (TABLE 1). Attrition was 15% and
behavior therapy group was asked to 22% at 3 and 6 months, respectively, and
submit self-monitoring diaries to the Dependent Measures did not vary by treatment group at ei-
therapist each week. All participants The primary dependent measure was ther assessment (3 mo, x2 =0.288, P=.59;
were contacted at 3 months and 6 change in body weight. Weight was mea- 6 mo, x2 =0.003, P=.96) (Figure 1). Par-
months to schedule individual appoint- sured in the clinic at baseline, 3, and 6 ticipants who did not attend the 6-month
ments for follow-up measurements and months in light street clothing, without follow-up were significantly younger
a brief 15 minute check-in with the shoes, and on a calibrated scale. Height (t=−2.75, P,.007) and had less e-mail
clinical psychologist. was measured using a wall-mounted sta- or Internet experience at baseline but did
diometer. The circumference at the waist not differ from attendees on baseline
Procedures for Internet (measured at the umbilicus) was mea- BMI, education, or level of depressive
Behavior Therapy sured with a Gulick steel tape measure symptoms.
Behavior therapy participants re- using the procedure recommended by
ceived all of the above plus the follow- Lohman et al.17 Physical activity was Changes in Body Weight
ing procedures. They were instructed measured at each assessment using a self- and Waist Circumference
to report self-monitoring information report format of the Paffenbarger activ- Analyses were conducted for the 65 par-
each week via an electronic diary ac- ity questionnaire.18 Dietary intake was ticipants with objective follow-up data
cessible on the study Web site. Weekly measured using the Block Food Fre- at all 3 assessments (33 behavior
self-monitoring information included quency Questionnaire19 at baseline, 3, therapy, 32 education). Repeated-
weight, calories, fat grams, and exer- and 6 months and was analyzed using the measures ANOVA examining weight
cise energy expenditure. Along with National Cancer Institute Dietary Analy- showed a significant treatment3time
submitting their diaries, participants sis System 4.01 software program. De- interaction (P=.005) (FIGURE 2). Those
were also able to submit any com- pressive symptoms were measured us- in the behavior therapy group lost more
ments or questions they had to the ing the Centers for Epidemiological weight than those in the education
therapist. Studies Depression Scale.20 Use of the group from baseline to 3 months. Both
1174 JAMA, March 7, 2001—Vol 285, No. 9 (Reprinted) ©2001 American Medical Association. All rights reserved.
a significant treatment 3time interac- *Values are expressed as No. (percentage) unless otherwise indicated.
waist circumference reduction was 3.0 group achieved the 5% weight loss goal 82
(4.0) cm at 3 months and 3.1 (4.4) cm (35% vs 18%; x2 =3.39; P=.07). 80
Mean Body Weight, kg
including all randomized participants us- (3.9) cm at 3 months and 2.3 (3.9) cm Sixty-five participants in this analysis completed all 3
ing baseline weight for anyone with by 6 months. In the behavior therapy assessments. (P= .005). Error bars indicate SEM.
missing data at any follow-up period. Re- group, mean (SD) waist reduction was
peated-measures ANOVA on the pat- 5.3 (4.9) and 4.6 (5.5), respectively. Post
tern of weight loss showed a significant hoc t tests showed that mean waist re- those in the education group (P,.001).
treatment 3time interaction (P,.001). ductions were significantly different be- Between months 3 and 6, logins de-
In the intent-to-treat analysis, the edu- tween the groups at both 3 months creased for both groups (P,.001); how-
cation group lost a mean (SD) of 1.0 (P=.001) and 6 months (P=.02). ever, participants in behavior therapy
(2.4) kg at 3 months and 1.3 (3.0) kg group still logged in more often during
by 6 months. Weight loss in the behav- Web Site Login Frequency this period—a mean (SD) of 6.8 (6.2)
ior therapy group was 3.2 (2.9) kg by 3 To obtain an objective measure of Web times compared with 1.0 (3.0) times
months and 2.9 (4.4) kg by 6 months. site use, login data for all participants among education participants (P,.001).
Post hoc t tests showed mean weight were tracked over the 24-week period. Login frequency was significantly cor-
losses were significantly different be- Participants in the behavior therapy related with weight change between 0
tween the groups at both 3 months group logged in to the Web site a mean and 6 months both in the behavior
(P,.001) and 6 months (P=.04). Within (SD) of 19 (10.9) times over the first 3 therapy (rs =−0.43; P=.003) and in the
the intent-to-treat sample, x2 tests were months compared with 8.5 (10.4) for education group (rs =−0.33, P=.03).
©2001 American Medical Association. All rights reserved. (Reprinted) JAMA, March 7, 2001—Vol 285, No. 9 1175
ity to detect differences in eating and ex- to Internet weight loss resources alone. motivationally-tailored physical activity intervention.
Ann Behav Med. 1998;20:174-180.
ercise between the groups may reflect Thus, the Internet appears to be a vi- 13. Taylor CB, Agras WS, Losch M, Plante TG, Bur-
difficulty of accurately measuring these able method for delivery of structured nett K. Improving the effectiveness of computer-
assisted weight loss. Behav Ther. 1991;22:229-236.
behaviors. It is also probable that par- behavioral weight loss programs deserv- 14. Burnett KF, Taylor CB, Agras WS. Ambulatory
ticipants in the behavior therapy group ing of future research. computer-assisted therapy for obesity: a new fron-
tier for behavior therapy. J Consult Clin Psychol. 1985;
became more accurate in their estima- 53:698-703.
Author Contributions:
tion of dietary intake and exercise be- Study concept and design: Tate, Wing, and Winett. 15. Taylor H. Harris Poll. Online population growth
cause they were self-monitoring intake Acquisition of data: Tate. surges to 56% of all. December 22, 1999. Poll No.
Analysis and interpretation of data: Tate, Wing, and 76. Available at: http://www.gsharrisinteractive.com
and activity; hence, there was a greater Winett. /harris_poll/index.asp?PID=9. Accessibility verified Feb-
association between behavior changes Drafting of the manuscript: Tate, Wing, and Winett. ruary 3, 2001.
Critical revision of the manuscript for important in- 16. Thomas S, Reading J, Shephard RJ. Revision of
and weight change in this group. tellectual content: Tate and Wing. the Physical Activity Readiness Questionnaire (PAR-Q).
The major strength of our study is that Statistical expertise: Tate and Wing. Can J Sports Sci. 1992;17:338-345.
Obtained funding: Tate and Wing. 17. Lohman TG, Roche AF, Martorell R, eds. Anthro-
it was a randomized trial with objective pometric Standardization Reference Manual. Cham-
Administrative, technical, or material support: Tate
weight and waist measurements and was and Wing. paign, Ill: Human Kinetics Books; 1988.
the first study to examine using Inter- Study supervision: Wing. 18. Paffenbarger RS, Wing AL, Hyde RT. Physical ac-
Funding/Support: This study was supported by a grant tivity as an index of heart attack risk in college alumni.
net technology to deliver a structured be- from the Weight Risk Investigators Study Council, a Am J Epidemiol. 1978;108:161-175.
havioral weight loss program. The pri- research division of Knoll Pharmaceutical. 19. Block G, Hartman AM, Dresser CM, Carroll MD,
Previous Presentation: Portions of this manuscript Gannon J, Gardner L. A data-based approach to diet
mary limitation of our study is that these formed the basis of Dr Tate’s doctoral dissertation and questionnaire design and testing. Am J Epidemiol.
results are for initial weight loss. The ef- have been presented previously at the North Ameri- 1986;124:453-469.
can Association for the Study of Obesity and Society 20. Radloff LS. The CES-D scale: a self-report depres-
ficacy of a program such as this for pro- of Behavioral Medicine annual scientific meetings. sion scale for research in the general population. Appl
ducing longer-term weight losses and Acknowledgment: We thank Don DeMaio for his Web Psychologic Measurement. 1977;1:385-401.
expertise and Sheila Jacobs from Lifespan Employee 21. Wing RR, Jeffery RW. Sample size in clinical out-
maintenance remains to be demon- come research: the case of behavioral weight con-
Health Promotion for her help with recruitment.
strated. Further studies with larger trol. Behav Ther. 1984;15:550-556.
samples including equal numbers of men 22. Thomas R, ed. Weighing the Options: Criteria for
REFERENCES Evaluating Weight Management Programs. Washing-
and women are needed to replicate these ton, DC: National Academy Press; 1995.
1. Flegal KM, Carroll MD, Kuczmarski RJ, Johnson CL.
findings. Furthermore, as this study was Overweight and obesity in the United States: preva-
23. Blackburn GL. Effects of weight loss on weight-
lence and trends. Int J Obes Relat Metab Disord. 1998; related risk factors. In: Brownell KD, Fairburn CG, eds.
an initial feasibility and efficacy trial, the Eating Disorders and Obesity: A Comprehensive Hand-
22:3947.
design does not allow for dismantling of 2. Sherwood NE, Morton N, Jeffery RW, French SA, Neu- book. New York, NY: Guilford Press; 1995:406-410.
24. Goldstein DJ. Beneficial health effects of modest
the behavior therapy program to deter- mark-Sztainer D, Falkner NH. Consumer preferences in
weight loss. Int J Obes Relat Metab Disord. 1992;16:
format and type of community-based weight control
mine the critical components of this in- programs. Am J Health Promot. 1998;13:12-18. 397-416.
tervention. Certainly, continued con- 3. Jeffery RW, Hellerstedt WL, Schmid TL. Correspon- 25. Miller WC, Eggert KE, Wallace JP, Lindeman AK,
dence programs for smoking cessation and weight con- Jastremski C. Successful weight loss in a self-taught,
tact has been a critical element of weight trol: a comparison of two strategies in the Minnesota Heart self-administered program. Int J Sports Med. 1993;
loss programs and other behavior- Health Program. Health Psychol. 1990;9:585-598. 14:401-405.
4. O’Loughlin J, Paradis G, Meshefedjian G, Kish- 26. Klem ML, Klesges RC. Competition in a minimal-
change programs. The difference in con- chuk N. Evaluation of an 8-week mailed health- contact weight-loss program. J Consult Clin Psy-
tact between conditions may be respon- weight intervention. Prev Med. 1998;27:288-295. chol. 1988;56:142-144.
5. Cameron R, MacDonald MA, Schlegel RP, et al. To- 27. Abrams DB, Follick MJ. Behavioral weight loss in-
sible for the differences in weight loss. tervention at the work site: feasibility and mainte-
ward the development of self-help health behaviour
Future studies could examine the ef- change programs: weight loss by correspondence. Can nance. J Consult Clin Psychol. 1983;51:226-233.
J Public Health. 1990;81:275-279. 28. Stunkard AJ, Brownell K. Work-site treatment for
fects of different types of contact on obesity. Am J Psychiatry. 1980;137:252-253.
6. Hellerstedt W, Jeffery R. The effects of a telephone-
weight loss using the Internet and e-mail. based intervention on weight loss. Am J Health Pro- 29. Jeffery RW, Danaher BG, Killen J, Farquhar JW,
Finally, it should be noted that partici- mot. 1997;11:177-182. Kinnier R. Self-administered programs for health be-
7. Meyers A, Graves T, Whelan J, Barclay D. An evau- havior change: smoking cessation and weight reduc-
pants in behavior therapy group had met lation of a television-delivered behavioral weight loss tion by mail. Addict Behav. 1982;7:57-63.
face-to-face with the therapist that they program: are the ratings acceptable? J Consult Clin 30. Jeffery RW, Gerber WM. Group and correspon-
Psychol. 1996;64:172-178. dence treatments for weight reduction used in the Mul-
had been corresponding with weekly by 8. Harvey-Berino J. Changing health behavior via tele- tiple Risk Factor Intervention Trial. Behav Ther. 1982;
e-mail. Other Internet programs using communications technology: using interactive televi- 13:24-30.
sion to treat obesity. Behav Ther. 1998;29:505-519. 31. Heska S, Greenway F, Anderson JW, et al. Self-
exclusively e-mail or online communi- 9. Winett RA, Moore JF, Wagner JL, et al. Altering help weight loss vs a structured commercial program
cations without an initial face-to-face shoppers’ supermarket purchases to fit nutritional after 26 weeks: a randomized controlled trial. Obes
meeting may produce different results. guidelines: an interactive information system. J Appl Res. 1999;7:19S.
Behav Anal. 1991;24:95-105. 32. Leermakers EA, Anglin K, Wing RR. Reducing post-
In summary, the results of this study 10. Winett RA, Anderson ES, Bickley PG, et al. Nu- partum weight retention through a correspondence
showed that a treatment program that trition for a Lifetime System(c): a multimedia system intervention. Int J Obes Relat Metab Disord. 1998;
for altering food supermarket shoppers’ purchases to 22:1103-1109.
included access to Internet weight loss meet nutritional guidelines. Comput Human Behav. 33. Jeffery RW, Wing RR, Thorson C, Burton LC. Use
resources, structured behavioral com- 1997;13:371-392. of personal trainers and financial incentives to in-
11. Prochaska JO, DiClemente CC, Velicer WF, Rossi crease exercise in a behavioral weight-loss program.
ponents, weekly contact, and individu- JS. Standardized, individualized, interactive, and per- J Consult Clin Psychol. 1998;66:777-783.
alized therapist feedback, delivered via sonalized self-help programs for smoking cessation. 34. Lichtman SW, Pisarska K, Berman ER, et al. Dis-
Health Psychol. 1993;12:399-405. crepancy between self-reported and actual caloric in-
e-mail, produced better initial weight 12. Marcus BH, Bock BC, Pinto BM, Forsyth LH, Rob- take and exercise in obese subjects. N Engl J Med.
losses compared with providing access erts MB, Traficante RM. Efficacy of an individualized 1992;327:1893-1898.
©2001 American Medical Association. All rights reserved. (Reprinted) JAMA, March 7, 2001—Vol 285, No. 9 1177