The Internet has potential as a medium for health behavior change programs, but no controlled studies have yet evaluated the impact of a fully automated physical activity intervention over several months with real-time objective feedback from a monitor.
The aim was to evaluate the impact of a physical activity program based on the Internet and mobile phone technology provided to individuals for 9 weeks.
A single-center, randomized, stratified controlled trial was conducted from September to December 2005 in Bedfordshire, United Kingdom, with 77 healthy adults whose mean age was 40.4 years (SD = 7.6) and mean body mass index was 26.3 (SD = 3.4). Participants were randomized to a test group that had access to an Internet and mobile phone–based physical activity program (n = 47) or to a control group (n = 30) that received no support. The test group received tailored solutions for perceived barriers, a schedule to plan weekly exercise sessions with mobile phone and email reminders, a message board to share their experiences with others, and feedback on their level of physical activity. Both groups were issued a wrist-worn accelerometer to monitor their level of physical activity; only the test group received real-time feedback via the Internet. The main outcome measures were accelerometer data and self-report of physical activity.
At the end of the study period, the test group reported a significantly greater increase over baseline than did the control group for perceived control (
A fully automated Internet and mobile phone–based motivation and action support system can significantly increase and maintain the level of physical activity in healthy adults.
Physical inactivity is a major concern for developed societies. It accounts for about 12% of all deaths [
Government organizations recognize physical activity, along with a healthy diet, as playing an important role in the prevention of obesity [
Internet-based behavioral change interventions minimize face-to-face interaction, thereby increasing cost-effectiveness through greater accessibility [
Internet-based physical activity websites differ in their level of interaction, from individually tailored assistance to general guidelines or advice [
We have developed a fully automated Internet, email, and mobile phone system [
Our primary hypothesis was that a group provided with access to the Internet and a mobile phone–based physical activity program would maintain a higher level of physical activity over 9 weeks than a control group who wore physical activity monitors but received no feedback and had no access.
A total of 140 people were initially recruited via a market research recruitment agency and passed the telephone screening (
All participants agreed not to take part in any other studies, were briefed on the study aims, and signed an informed consent form in accordance with the Declaration of Helsinki [
Flow diagram of study participation
The 77 participants came to the center and were issued a wrist-worn accelerometer and Bluetooth-compatible mobile phone (Nokia 6230, with their SIM card inserted). After 3 weeks of monitoring baseline physical activity, participants returned and were stratified by age, gender, and BMI and were randomly allocated to either the control (n = 30) or test group (n = 47). More participants were allocated to the test group in order to maximize information on use of the system. All participants received £30 for attending the initial screening at the center, £140 to cover mobile phone costs, and £290 at closeout.
Although pedometers are low cost, they are typically attached to a waist band and therefore primarily record walking, making 24-hour monitoring more difficult. In contrast, accelerometry tools record a wider range of movement and have greater flexibility for body positioning, allowing for sustained monitoring even during sleep. Accelerometers have been widely used to monitor physical activity [
Together with a technology company [
The Internet, email, and mobile phone behavior change system (
Behaviour change system home page
A text-based automated dialogue module helped participants identify their perceived barriers and offered tailored solutions (see the Multimedia Appendix). For example, for the barrier “You can’t exercise because there’s something else on at the same time,” one of the solutions offered could be “Form a habit: If you always exercise on the same days at the same time, your routine will become a fixture in your life, not a whim. Not exercising will feel unnatural. Choose some days of the week where you’ll always reserve a slot for exercise, starting now!” Solutions were tailored to the individual via an underlying matrix that contained a strength of association between each barrier and solution. The strength of association between solutions and barriers increased in line with the increase in the level of physical activity of participants who had previously selected them.
Participants were also encouraged to select three motivating benefits, for which email and/or mobile phone text messages were optional. There was a library of information on a range of different physical activities, from household duties to team sports, and a chat-room style message board. Charts displayed real-time output from their physical activity monitor in three bands, moderate, high, and very high, with summaries for that day, the week (
Weekly activity charts
At the study center, participants received a full explanation about all procedures and were given an opportunity to ask questions. They were instructed to wear the Bluetooth Actiwatch on the wrist of their nondominant arm continuously for the following 12 weeks (3 weeks baseline and 9 weeks intervention). As the accelerometer was not fully waterproof, participants were asked to remove it when washing, bathing, or swimming. Following collection of 3 weeks of baseline data, the test group participants received a short demonstration of the Internet-based behavior change system; the control group also came to the center but only received verbal advice on recommended physical activity levels. The test group then had access to the Internet-based behavior change system for 9 weeks, whereas the control group had no access and received no feedback.
The primary dependent measure was change in moderate physical activity recorded by the longer version of the International Physical Activity Questionnaire (IPAQ) [
A set of cognitive items was developed specifically for the study, each scored against a 7-point numbered scale ranging from “Strongly Disagree” to “Strongly Agree.” The items to measure motivation were as follows: “I am very satisfied with my level of fitness,” “I am very satisfied with my current level of motivation to exercise,” “I consider myself to be very healthy,” and “I am very happy about my general level of well-being” (Cronbach alpha = .89). The items to measure perceived control were “Exercise is too much effort” and “I feel in control of how much exercise I get” (Cronbach alpha = .63), and those measuring intention/expectation to exercise were “I intend to exercise for 30 minutes at least 3 times in the next week” and “I realistically expect that I will actually exercise for 30 minutes at least 3 times in the next week” (Cronbach alpha = .92). One item measured participant interest in using an Internet-based behavior change system: “I think an Internet-based motivation program could help people to take more exercise.”
Participants also completed an exercise Skills and Knowledge Questionnaire that asked about skills used to increase physical activity [
Three participants were found to have faulty Actiwatches and so were removed from all statistical analyses. IPAQ data were processed according to the Guidelines for Data Processing and Analysis of the International Physical Activity Questionnaire [
Actiwatch data for contiguous periods of zero extending longer than 30 min were omitted, as were data for periods with > 5000 counts for longer than 10 min since both these conditions indicated temporary malfunctioning of the accelerometer. Only days that had at least 10 h of recorded data following these corrections were retained for analysis [
A Generalized Estimating Equation Model with log link and Poisson distribution was used to calculate the number of 2-min epochs spent within three metabolic equivalent (MET) ranges [
We focused on the difference in total time of nonsedentary physical activity between the two groups rather than the absolute amount of physical activity within each group, as estimates of the latter can vary by a factor of 10 depending on the threshold point used [
Participants were instructed to remove the accelerometer for swimming—an activity selected by 36% of the test group who logged on. Therefore, our accelerometer-based estimate of physical activity did not fully account for all exercise undertaken, potentially attenuating any differences observed between the test and control groups. Anthropometric measures at the end of the study were analyzed using normal analysis of covariance models with baseline prestudy values as covariate. All analysis was carried out using SAS, version 9.1.3 [
A preliminary analysis showed that there were no differences between groups for baseline measures of age, weight, BMI, percent body fat, blood pressure, or initial level of physical activity, whether measured by the Actiwatch or IPAQ (
Baseline characteristics of participants
|
|
|
|
Women (%) | 64 | 70 | .63 |
White ethnicity (%) | 100 | 97 | .39 |
Household broadband access (%) | 29 | 22 | .43 |
Age (years) | 40.5 (7.1) | 40.1 (7.7) | .97 |
Weight (kg) | 75.1 (11.7) | 73.9 (10.2) | .60 |
Height (cm) | 166.3 (6.6) | 165.2 (7.7) | .38 |
BMI | 26.2 (2.8) | 26.5 (4.1) | .68 |
Percent body fat (%) | 30.2 (6.5) | 31.0 (10.1) | .52 |
Blood pressure (mmHg) Systolic | 119.8 (7.7) | 118.2 (8.4) | .40 |
Blood pressure (mmHg) Diastolic | 78.3 (5.7) | 77.9 (6.1) | .82 |
Actiwatch accelerometer-measured time (epochs) spent above 3 and up to 6 METs during 3-week initiation period | 228.0 (52.1) | 214 .2 (53.1) | .11 |
Initial IPAQ self-report level of physical activity (MET mins) | 4350 (3200) | 3868 (2257) | .44 |
*Values are expressed as mean (SD) except for the first three variables.
†
More than 85% (mean = 86.4%, SD = 2.1) of test participants logged on each week during the first 4 weeks, decreasing to a plateau around 75% (mean = 76.1%, SD = 5.1) for the last 5 weeks. This level is lower than for partially automated behavior change systems [
The average number of log-ons per week was 2.9 (SD = 0.5), with short average duration of 7.5 min (SD = 0.9). The most frequently used components of the system were the activity charts (showing the accelerometer feedback data), the schedule (weekly exercise planner), and chat-room style message board. All components of the system were accessed by at least 33% of the participants during the intervention period. Typically, participants quickly formed an idiosyncratic preference for a few components of the system and then repeatedly used these throughout the intervention.
Comments on the message board indicated that participants found the system both educational and motivational, for example, “I am amazed looking at the graphs sometimes — I took my little fella to Bezerks in Northampton on Thursday morning and the graph went crazy with all the running around I did!”
The most popular (frequent) benefits of exercise were “Exercise will help me with weight loss” (n = 19), “I will have more energy” (n = 13), and “I will improve my muscle tone” (n = 11). The most commonly selected barriers to physical activity were time conflicts (n = 27), low motivation (n = 11), and procrastinating (n = 6).
As shown in
Self-reported physical activity in test and control groups
|
|
|
|
|
Overall | 12.0 (3.1) | 4.0 (4.1) | .12 | −2.3 to 18.3 |
Leisure time | 4.1 (2.6) | −5.5 (3.5) | .03 | 0.8 to 18.3 |
Overall | −5.9 (2.0) | 1.4 (2.7) | .03 | −14.0 to −0.5 |
Weekday | −5.2 (1.7) | −0.2 (2.3) | .08 | −10.8 to 0.7 |
Weekend | −0.9 (0.6) | 1.2 (0.8) | .04 | −4.2 to −0.1 |
*Values expressed as mean (SD).
†
We collected 4811124 data points from study participants, which represented 94.1% of the total expected based on the first and last recorded points for each individual (5114431).
Average sleep times for the two groups were not significantly different (test group: 467 min, SD = 40; control group: 468 min, SD = 38;
Accelerometer-measured mean number of 2-min epochs spent in moderate intensity MET range (above 3 and up to 6); error bars represent SE; baseline is 3-week average before start of intervention
At the end of the study period, the test group reported a significantly greater increase over baseline than did the control group for the perceived control factor (mean change test group = +0.57; mean change control = −0.37;
The Skills and Knowledge Questionnaire indicated that, after adjusting for baseline, the test group had a significantly higher sense of internal control (test mean = 7.24; control mean = 5.85;
After the study period, the test group had a significantly greater interest in using an Internet-based behavior change system than the control group (test group = 4.92; control = 3.85;
The difference between the change in the test group’s BMI (mean change = −0.24, SE = 0.11) over the study period and that for the control group (mean change = 0.10, SE = 0.14) approached significance (
Increasing physical activity in the general population has an important role in the prevention of obesity and associated health problems [
Although we observed an increase in accelerometer-measured physical activity for the test group over the control group, our analysis was limited by its uniaxial nature; future studies could employ a triaxial accelerometer so that greater differentiation of physical activity types can be achieved [
The difference between the test and control group accelerometer-measured physical activity was apparent for most of the 9-week intervention (see
In line with the Theory of Planned Behavior [
It was clear from the verbatim comments posted by participants on the message board that the accelerometer-based activity charts acted as educational information, allowing them to link periods of high physical activity to events in their everyday life. Indeed, everyday physical activities such as walking are considered to have the greatest potential for increasing overall activity levels of a sedentary population [
In line with other research [
Most participants selected from a relatively small set of barriers and were motivated by similar benefits, as has been reported by other researchers [
Based on a range of behavior change principles taken from the literature [
In summary, we found that participants with access to a fully automated behavior change system engaged in, on average, 2 h 18 min more physical activity per week than those with no access.
Jaspreet Singh Sodhi had full access to all of the data and takes responsibility for the integrity of the data and the accuracy of the data analysis.
None declared.
Screenshots of the behaviour change system (ppt)
body mass index
International Physical Activity Questionnaire
metabolic equivalent