This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Evaluation of online health interventions should investigate the function of theoretical mechanisms of behavior change in this new milieu.
To expand our understanding of how Web-based interventions influence behavior, we examined how changes at 6 months in participants’ psychosocial characteristics contributed to improvements at 16 months in nutrition, physical activity (PA), and weight management as a result of the online, social cognitive theory (SCT)-based Guide to Health intervention (WB-GTH).
We conducted recruitment, enrollment, and assessments online with 272 of 655 (41.5%) participants enrolling in WB-GTH who also completed 6- and 16-month follow-up assessments. Participants’ mean age was 43.68 years, 86% were female, 92% were white, mean education was 17.45 years, median income was US $85,000, 84% were overweight or obese, and 73% were inactive. Participants received one of two equally effective versions of WB-GTH. Structural equation analysis of theoretical models evaluated whether psychosocial constructs targeted by WB-GTH contributed to observed health behavior changes.
The longitudinal model provided good fit to the data (root mean square error of approximation <.05). Participants’ weight loss at 16 months was predicted by improvements in their PA (betatotal = -.34,
The WB-GTH influenced behavior and weight loss in a manner largely consistent with SCT. Improving social support, self-efficacy, outcome expectations, and self-regulation, in varying combinations, led to healthier diet and exercise habits and concomitant weight loss. High initial levels of self-efficacy may be characteristic of Web-health users interested in online interventions and may alter the function of SCT in these programs. Researchers may find that, although increased self-efficacy enhances program outcomes, participants whose self-efficacy is tempered by online interventions may still benefit.
Clinicaltrials.gov NCT00128570; http://clinicaltrials.gov/ct2/show/NCT00128570 (Archived by WebCite at http://www.webcitation.org/5vgcygBII)
As many as 79% of Americans use the Internet, with growing majorities across racial, ethnic, educational, and income groups reporting going online – 73% go online daily. The vast majority of users go to the Internet for information on health topics [
Randomized control trials of Internet-based interventions have largely produced modest, short-term effects [
In addition to limited evidence of long-term effectiveness, attrition is typically high in Internet-based trials (43%-50%) [
The reach of entirely online interventions is defined as the percentage and representativeness of individuals willing to participate [
Finally, Internet-based health promotion interventions should be based on theory and evaluated in a way to validate and refine the application of theory within the new delivery environment [
Social cognitive theory of health behavior
The current study examined the relationship of SCT variables to behavior and weight change in the Web-Based Guide to Health intervention (WB-GTH). An earlier site-based version of GTH suggested that self-efficacy, self-regulation, and social support (and, to a lesser extent, outcome expectations) can mediate the effects of SCT-based health interventions in a manner consistent with theory [
The WB-GTH featured online recruitment, screening, consent, assessment, and program implementation with and without enhanced self-regulatory components (basic WB-GTH and enhanced WB-GTH versions). Recruitment into the WB-GTH, described in detail elsewhere [
The purpose of the current study was to determine whether improvements at 6 months in social support, self-efficacy, outcome expectations, and self-regulation preceded observed behavior and weight change in a manner consistent with SCT [
The WB-GTH trial was conducted entirely online (baseline assessment from September 2007 to November 2008; 6-month assessment from April 2008 to May 2009, 16-month assessment from February 2009 to March 2010). Of 655 randomly assigned participants, 199 quickly withdrew from the program, logging in to only one or two program modules. Participants who quickly withdrew from WB-GTH were less active, but otherwise did not differ in demographics or behavior from those who engaged with one of the two versions of the WB-GTH (ie, saw at least three program modules; n = 456; [
The WB-GTH program (described in detail elsewhere [
The basic WB-GTH program, used by 51.1% (139/272) of current study participants, provided generic goals for adding steps and minutes of walking to their daily routines (ie, add 400 steps/day each week up to 3000 steps and 5 minutes/day up to 30 minutes, 5 days/week). Other goals included adding fitness walking after reaching 30 minutes of walking 5 days a week, adding fruits and vegetables (F&V) (1/day each week to reach 5-9 for female and 5-10 for male users), adding whole-grain foods and low-fat dairy foods (1/day each week for up to 3 servings a day), and decreasing high-fat and high-sugar foods (no more than 28 servings/week). Participants kept and reported daily logs of steps, minutes walked, enjoyment of and exertion during planned walks, and intake of fruits, vegetables, whole grains, low-fat dairy, and high-fat and high-sugar foods, and they weighed themselves weekly. Each week users of the basic WB-GTH program received general feedback (eg, comparison of current levels and overall goals) with a restatement of generic goals and a prompt for users to plan for the next week. The enhanced WB-GTH program, used by 48.9% (133/272), differed from the basic program by providing users with personalized feedback and tailored goal setting and planning, and by allowing participants to select, report, and receive feedback on specific behavior-change strategies for increasing PA and improving nutrition [
Participants completed the Block 2005 Food Frequency Questionnaire (NutritionQuest, Berkeley, CA, USA) [
Participants used a pedometer (Digi-walker SW-200, Yamax USA, Inc, San Antonio, TX, USA) and a digital bathroom scale (Health-o-meter HDL150-01, Sunbeam Products, Inc, Maitland, FL, or Tanita HD-313, Tanita Corporation, Arlington Heights, IL, USA) provided at baseline. Participants wore their pedometers for 7 days and logged the number of steps registered each day. They returned to the WB-GTH website to report daily steps for at least 4 consecutive days. Participants also completed an online questionnaire about the duration, pace, and number of walking, treadmill, jogging, and running sessions they took in a typical week. The metabolic equivalent (MET; ie, the ratio of work metabolic rate to a standard resting metabolic) for each activity was computed and summed across activities to compute walking METhours/week for each participant at baseline and 16 months. Participants used the bathroom scale provided to measure body weight in pounds without clothing just after waking in the morning or before going to bed in the evening. They reported their weight and height online along with their walking log data. We examined mean daily steps (total steps reported divided by days of pedometer use), mean walking METhours/week, body weight in pounds, and change in these variables during the 16-month interval.
The Health Beliefs Survey administered online at baseline and 6 months measured change in nutrition- and PA-related social support, self-efficacy, outcome expectations, and self-regulation (see
Multivariate repeated measures analysis of variance (MANOVA) evaluated effects of the WB-GTH on SCT variables at 6 months, and nutrition, PA, and body-weight variables at 16 months. Full information maximum likelihood (FIML) latent-variable structural equation modeling (LISREL version 8.81, Scientific Software International, Inc, Lincolnwood, IL, USA) [
Health Beliefs Survey: scale descriptions and internal consistency estimates
Variable description | Subscale | Number of items | Alphaa | |
|
||||
Social support | Family | 11 | .90 | |
Friends | 11 | .94 | ||
Self-efficacy | Eating healthy foods | 12 | .91 | |
Avoiding high-fat and high-sugar foods | 15 | .90 | ||
Planning and tracking intake | 10 | .96 | ||
Positive physical and self-evaluative outcome expectations | 10 | .90 | ||
Negative physical, social and self-evaluative outcome expectations | 11 | .89 | ||
Self-regulation | Planning and tracking | 11 | .92 | |
High-fat and high-sugar foods | 13 | .90 | ||
Healthy food choices | 8 | .90 | ||
|
||||
Social support | Family | 8 | .94 | |
Friends | 8 | .96 | ||
Self-efficacy | 23 | .95 | ||
Positive outcome expectations | Physical | 5 | .83 | |
Affective | 5 | .66 | ||
Negative outcome expectations: physical, social, and self-evaluative | 7 | .88 | ||
Self-regulation | Setting goals and planning activity | 9 | .91 | |
Tracking physical activity | 5 | .85 | ||
Increasing enjoyment | 3 | .77 |
a Cronbach alpha coefficient of internal consistency.
Participants viewed a mean of 21.33 (SD 17.19) modules: 18% (49/272) viewed only introductory WB-GTH modules (ie, modules 1-5), 37.1% (101/272) also viewed WB-GTH core-content modules (6-16), and the remaining 44.8% (122/272) viewed maintenance modules (>17 modules). Participants viewed similar numbers of modules of each version: basic WB-GTH (mean 22.36, SD 17.51) and the enhanced WB-GTH (mean 20.34, SD 16.87; F1,270 = 0.95,
MANOVA of baseline and 16-month data indicated that WB-GTH participants viewing both versions of WB-GTH made behavioral and weight changes (time: F8,157 = 10.02, partial-eta squared = .34,
Participants’ improvements in behavior and weight at 16 months were preceded by changes at 6 months in social cognitive characteristics related to nutrition (time: F11,235 = 42.91, partial-eta squared = .67,
Baseline, 6-month, and 16-month follow-up weight, physical activity, and nutrition-related outcomes of users of the Web-Based Guide to Health intervention
Outcome | Study time pointa | Mean | SD | F | dfb |
|
Partial-eta squaredc | |
|
base | 176.98 | 28.03 | 29.01 | 1,194 | <.001 | .130 | |
16 mo | 172.34 | 28.01 | ||||||
|
||||||||
Fruit servings/day | base | 1.13 | 0.81 | 83.34 | 1,236 | <.001 | .261 | |
16 mo | 1.66 | 0.95 | ||||||
Vegetable servings/day | base | 2.97 | 1.93 | 55.128 | 1,236 | <.001 | .189 | |
16 mo | 3.95 | 2.36 | ||||||
Percentage kilocalories from fat | base | 36.62 | 5.86 | 37.33 | 1,236 | 0.00 | .14 | |
16 mo | 34.40 | 6.21 | ||||||
Percentage kilocalories from sweets | base | 14.53 | 9.32 | 45.07 | 1,236 | 0.00 | .16 | |
16 mo | 11.30 | 8.50 | ||||||
Kilocalories/day | base | 1820.95 | 700.81 | 22.36 | 1,236 | 0.00 | .09 | |
16 mo | 1641.14 | 510.47 | ||||||
Family social support | base | 2.70 | .81 | 57.193 | 1,245 | <.001 | .189 | |
6 mo | 3.07 | .83 | ||||||
Friend social support | base | 2.91 | .78 | 29.768 | 1,245 | <.001 | .108 | |
6 mo | 3.16 | .74 | ||||||
Self-efficacy for avoiding fat and sugar | base | 75.99 | 17.29 | 2.875 | 1,245 | .091 | .012 | |
6 mo | 77.59 | 16.81 | ||||||
Self-efficacy for eating healthy foods | base | 73.48 | 16.80 | 3.794 | 1,245 | .053 | .015 | |
6 mo | 75.59 | 17.73 | ||||||
Positive outcome expectations | base | 4.32 | .61 | 0.01 | 1,245 | 0.95 | .00 | |
6 mo | 4.32 | .57 | ||||||
Negative outcome expectations | base | 2.90 | .73 | 16.89 | 1,245 | 0.00 | .06 | |
6 mo | 2.72 | .81 | ||||||
Self-regulation healthy food choices | base | 3.11 | .82 | 301.23 | 1,245 | 0.00 | .55 | |
6 mo | 4.06 | .62 | ||||||
Self-regulation high-fat/high-sugar foods | base | 3.01 | .75 | 292.27 | 1,245 | 0.00 | .54 | |
6 mo | 3.84 | .61 | ||||||
Self-regulation planning and tracking nutrition | base | 2.35 | .81 | 376.72 | 1,245 | 0.00 | .61 | |
6 mo | 3.58 | .79 | ||||||
|
||||||||
Steps/day | base | 6252.89 | 1876.34 | 40.32 | 1,177 | 0.00 | .19 | |
16 mo | 7741.47 | 3247.56 | ||||||
METhours/week walkingd | base | 2.82 | 9.70 | 45.20 | 1,177 | 0.00 | .20 | |
16 mo | 12.43 | 18.02 | ||||||
Family social support | base | 2.49 | 1.03 | 19.98 | 1,245 | 0.00 | .08 | |
6 mo | 2.77 | 1.01 | ||||||
Friend social support | base | 2.88 | .98 | 8.05 | 1,245 | 0.01 | .03 | |
6 mo | 3.07 | .87 | ||||||
Self-efficacy | base | 64.09 | 19.24 | 4.06 | 1,245 | 0.05 | .02 | |
6 mo | 61.28 | 22.24 | ||||||
Positive affective outcome expectations | base | 16.24 | 3.92 | 0.23 | 1,245 | 0.63 | .00 | |
6 mo | 16.38 | 4.57 | ||||||
Positive physical outcome expectations | base | 20.49 | 4.22 | 2.33 | 1,245 | 0.13 | .01 | |
6 mo | 20.05 | 4.81 | ||||||
Negative outcome expectations | base | 10.61 | 4.84 | 0.01 | 1,245 | 0.93 | .00 | |
6 mo | 10.65 | 5.13 | ||||||
Self-regulation goal setting and planning | base | 1.98 | .76 | 439.28 | 1,245 | 0.00 | .64 | |
6 mo | 3.47 | .94 | ||||||
Self-regulation tracking activity | base | 1.53 | .71 | 604.49 | 1,245 | 0.00 | .71 | |
6 mo | 3.65 | 1.19 | ||||||
Self-regulation increase enjoyment | base | 1.79 | .84 | 387.59 | 1,245 | 0.00 | .61 | |
6 mo | 3.30 | 1.04 |
a Baseline (base), 6 months (6 mo), or 16 months (16 mo).
b df: degrees of freedom for F test result.
c partial-eta squared.
d MET: metabolic equivalent.
A longitudinal, latent variable, structural model incorporated data from SCT variables exhibiting significant change (see
Social cognitive model of behavior and weight change among users of the Web-based Guide to Health intervention. F&V: fruits and vegetables; NOE: negative outcome expectations; PA: physical activity; SE: self-efficacy; SR: self-regulation; SS: social support. a
Completely standardized significant direct effect coefficients resulting from the structural analysis appear next to their associated paths; the PA portion of the model is at the top of
The nutrition portion of the model is at the bottom of
The model also evaluated whether participants’ weight loss at 16 months related to concomitant changes in nutrition and PA or to psychosocial changes that preceded dietary and PA improvements. The model explained 20% of the variance of change in weight. Participants’ weight loss at 16 months was related to increases in PA (betatotal = -.34,
The WB-GTH is an Internet intervention designed to improve nutrition and PA and prevent further weight gain in overweight to obese, inactive, but otherwise healthy adults. Based on SCT, the WB-GTH was delivered with and without enhanced self-regulatory features to Internet users who were recruited, screened, asked for consent, and assessed entirely online. The 16-month outcomes suggest that WB-GTH users lost almost 5 pounds (~3%) of body weight; increased daily step counts by 24%; made fourfold increases in weekly METS expended in walking; decreased calories from fat by 2%, calories from sweets by 3%, and daily calories by 10%; and increased F&V intake by about 1.5 servings a day. These changes were consistent with those exhibited by users of a site-based version of GTH delivered with enhanced supports [
A presumed strength of Internet interventions is that they can be highly flexible and personalized based on individual participant data [
The WB-GTH, like many other Internet programs, ultimately attracted a relatively affluent, predominantly female and white sample [
Discussed in detail elsewhere [
Finally, although an earlier version of GTH was shown to be effective compared to an untreated control condition [
The longitudinal design of the study strengthened the structural equation analysis, but the ordering of the SCT variables was theoretical and not chronological [
Although the current study demonstrates how SCT variables can contribute to the effectiveness of online interventions such as WB-GTH, researchers may need to go beyond traditional cognitive and motivational variables in order to explain a larger proportion of behavior change – perhaps, for example, assessing the affective and selective processes delineated in SCT that are associated with behavior change [
The outcomes of WB-GTH suggest that purely Web-based interventions can operate in a manner consistent with underlying theory. Further, despite problems with attrition and a relatively homogeneous reach, Internet programs such as WB-GTH can be effective in guiding users to adopt healthier nutrition, PA, and weight-management habits, which rival changes produced by more intensive, face-to-face interventions.
We thank Ashley Dorough, MS for her involvement in piloting and refining the physical activity component of the intervention and Todd Bowden and Shane Moore of PC Resources, Inc. for implementing the web-based components of the study.
None declared
fruits and vegetables
full information maximum likelihood
multivariate repeated measures analysis of variance
metabolic equivalent
negative outcome expectations
physical activity
root mean square error of approximation
social cognitive theory
self-efficacy
self-regulation
social support
Web-Based Guide to Health intervention