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.
Cost-effective interventions to improve diet and physical activity are a public health priority.
(1) To describe the components and behavioral principles underlying
The
Participants were 19 to 65 years (mean 44.0 +/- 10.6), and 74.3% (585/787) were female. Mean SF-8 Physical quality of life score increased significantly more in the intervention group than in the control group, 1.84 (95% CI 0.96-2.72) vs 0.72 (95% CI -0.15-1.58) respectively,
Cost-effective methods that can reach large populations with science-based interventions are urgently needed.
Clinicaltrials.gov NCT00607009; http://clinicaltrials.gov/ct2/show/NCT00607009 (Archived by WebCite at http://www.webcitation.org/5cLpCWcT6)
The important role of diet and physical activity in reducing the burden of chronic disease and obesity is well-established [
Despite the substantial evidence linking these behaviors to health outcomes, the great majority of Americans do not meet dietary and physical activity guidelines [
Intervention programs can change these behaviors, and a number of on-site and face-to-face programs have been found to be effective, such as those described by Beresford et al [
A number of research groups have developed effective mailed or computer-based and computer-tailored interventions, including Campbell et al [
The primary outcomes of the randomized controlled trial were change in diet and physical activity. Those results are in preparation [
Here we describe the components and principles of the
Diet and physical activity health risk assessments (HRAs), described in more detail below, are delivered via email and take approximately 15 minutes to complete.
The physical activity questionnaire was adapted from the Cross-Cultural Activity Participation Study (CAPS) questionnaire [
The dietary questionnaire contains 35 items, asks about “usual” intake, and includes both frequency and portion size. Foods were identified for inclusion based on analyses of the National Health and Nutrition Examination Survey (NHANES) 1999-2004 [
A second questionnaire, again delivered via email, obtains demographic data, tailoring information, and information related to assessing secondary outcomes. Tailoring information includes presence of children at home, habits related to cooking and eating out, physical activity preferences such as structured, facility-based exercise or lifestyle physical activity, and stage of readiness for change [
In this questionnaire, participants identify barriers that may get in the way of achieving their health behavior goals. Subsequent messages provide tips for overcoming their reported barriers.
Feedback is provided immediately after the participant submits the HRA. Separate reports are made of the participant’s intake of saturated fat, trans fats, added sugars, fruits and vegetables, and amount of physical activity, in relation to national and international guidelines [
After receiving the feedback, individuals may choose to participate in the full
Each week, the participants receive an email suggesting four to six small-step goals which are tailored to the individual characteristics mentioned above (
Participants are asked to commit to one or two of these to work on for the following week. The purpose of the tailoring is to identify small-step goals that are relevant to the individual participant and that take into account his or her constraints and preferences. These are small achievable goals, such as “I will have a salad with lunch two days this week” or “I will walk 20 minutes at lunch time”. Dietary goals are also suggested based on the individual’s reported intake. For example, a person who eats doughnuts twice a week may receive a suggested goal to eat them only once a week, or to eat a smaller portion. Physical activity goals are also tailored to a combination of stage of change and initial level of activity: persons reporting precontemplation or low/no activity will initially be given goals that facilitate their getting started, such as easy walking or buying walking shoes.
In subsequent weeks, in the email delivering the next set of goals, the participant is also asked whether or not the previous week’s goals were achieved. This is recorded in the Goal Tracker (see below).
Illustrative tailoring characteristics and associated suggested goals
Characteristic of Participant | Sample Small-Step Goal |
|
|
Early stage, prefers lifestyle activities, no children at home | I will make a date with a friend to go for a walk instead of for coffee or a drink. |
Early stage, has children at home | I will go to the playground with my kids two days this week after school/work, and walk around the playground. |
Action stage, prefers exercise activities, has children at home | I will get a family fitness video or DVD and do it with my kids at least one day this week. |
Action stage, prefers lifestyle activities, no children at home | I will walk to do errands or window shop on my lunch hour rather than sitting in cafeteria or at my desk, at least two days this week. |
|
|
Most dinners eaten at home, participant does the cooking | This week I will buy olive oil, and use it when I fry or stir-fry |
Eats out frequently | I will look for opportunities to eat whole grain foods when I eat out this week. |
Conditional (eats sweetened cereal) | This week when I shop, I will read the label on the box, and choose a cereal with less sugar. |
Conditional (eats sweetened cereal) and has children at home | This week when I shop, I will show my children how to read the label, and choose a cereal with less sugar. |
|
|
Eats out frequently | I will add vegetables to pizza or other carry-out this week. |
Most dinners eaten at home, no children at home | I will try to eat one new fruit and one new vegetable this week (different from what I usually eat). |
Most dinners eaten at home, children at home | I will have the kids participate in grocery shopping this week and choose one vegetable or fruit they are willing to eat. |
Participant does the cooking | On two days this week, I will build vegetables into the main dish, like adding frozen green beans to stew. |
Example of weekly email
A brief email mid-week reminds the participants of the goals they have chosen.
Immediately after choosing a goal, the participant is taken to his or her “personal home page” containing tips for achieving the goal(s) they have chosen, tips regarding the barriers they mentioned, a goal tracker, an interactive simulation tool, health information, and links to sites for additional information, such as government and organizational websites. Thus, the act of choosing a goal in the email reader ensures that 100% of participants who choose a small-step goal for the week will also view the additional home page content described below; no additional initiative on the part of the participant is required. See
Each week, participants receive tips on ways to achieve the specific small-step goals they have chosen that week; tips are also tailored to the factors above. They also receive tips on how to handle specific barriers that the participant has reported as constraints, such as time, money, or travel.
The program tracks which goals the participant has successfully achieved and categorizes them as to type of goal (eg, change in frequency vs change in amount). This is available on the participant’s personal home page. This information was not used in the evaluation of the effectiveness of the program, but was provided as an aid to the participant in understanding what types of goals work for that individual.
The simulation tool is an interactive feature of the
Each week, a different topic relevant to the selected intervention objective (ie, physical activity, fruits/vegetables, or carbs/fats) is discussed in a “Health Note”. Topics include research on the relation of physical activity, fruits and vegetables, or saturated and trans fats to heart disease, healthy weight, various cancers, metabolic syndrome, mental health, and cognitive decline. Knowledge relevant to the particular intervention objective is also provided, such as the components of fitness, trends in physical activity, and different types of fats. A brief summary of the topic appears in each weekly email, and the full article is presented on the individual’s personal home page.
Weekly suggested goals and tips promote building social support by suggestions such as walks with colleagues at lunch time. Equally important,
The behavioral strategies underlying
These behavioral strategies are applied in a basic structure of bringing forth a desirable behavior and providing the cues and repetition that help make the new behaviors habitual (
The Alive! behavior change model
A randomized controlled trial was conducted among non-medical regional employees of Kaiser Permanente of Northern California (KP). Persons employed in the Kaiser Division of Research, of which Dr. Sternfeld is a member, were not eligible to participate. Recruitment began in July 2006 and was accomplished in approximately three weeks. The intervention and follow-up was completed in December 2006 (
Randomized controlled trial intervention and follow-up
Employees were recruited through an invitational email sent from KP administrative offices, which included the diet and physical activity questionnaires described above. All employees were eligible. There was no monetary incentive to participate in the assessment or the subsequent randomized trial. The number who took the assessments and received individualized feedback but did not choose to join the randomized trial was not tracked.
Persons who agreed to participate in the randomized trial were automatically randomized by the program to either the intervention group or the control group. Randomization was by department (n = 192 departments) after stratification by department size, using a random number table. The control group was a delayed control, and control group participants were offered the program 8 months after the initial randomization. Thus, participants were aware of their randomization group. The delivery of the intervention was completely automated and did not involve any investigator actions. Diet and physical activity questionnaires and questionnaires on secondary outcomes like health status were automatically administered by the program at baseline and at the conclusion of the intervention.
After randomization, participants in the intervention group chose the intervention path they wanted to pursue: Physical Activity (PA); Fruits and Vegetables (FV); Fats and Sugars (FS). Participants received intervention messages only for the chosen path. Neither participation in
Data for secondary outcomes include self-assessed health status and health-related quality of life, using the SF-8 Health Survey questions [
The SF-8 Health Survey [
Presenteeism [
Self-efficacy [
Stage of Readiness for Change [
Results were analyzed by strict intention to treat, in which persons who did not respond to the follow-up questionnaire and therefore had missing data are included in the intention-to-treat analysis and assigned a change score of zero (119 of 436 in the control group, 27.3%; and 119 of 351 in the intervention group, 33.9%). Ordinal logistic regression models (Proc Genmod, SAS Institute, Cary, NC) were used for analyses of ordinal change variables. Multiple linear regression models (Proc Mixed, SAS Institute, Cary, NC) were used for analyses of the change in SF-8 quality of life. In all models, change in behavior was the dependent variable, randomization group was the primary fixed effect, department was a random effect factor, and all models were adjusted for age, gender, ethnicity, and baseline value of the dependent variable. Results are presented for the overall comparison of the intervention and control groups, although it should be noted that participants only received messages and goals relevant to the specific chosen path (Physical Activity, Fruits/vegetables, Fats/Sugars). Results presented as intention-to-treat probably represent an underestimate of effects, since they include all randomized participants including non-responders to the follow-up questionnaire (these were deemed to have a change score of zero, even though some of the non-responders may have experienced improvements in these behaviors).
The trial includes 787 persons who gave informed consent to be randomized. A larger number completed the assessments and received feedback, but that number is not tracked by the system. Of the 787 participants in the trial, 351 (45%) were randomized to the intervention group and 436 (55%) to the control group. The mean age was 44 years (range 19-65 years), 202 (25.7%) were men, and 70% had a college degree or higher education (
Demographic characteristics by treatment group and by intervention path (Alive! randomized trial, Oakland, CA 2006)
Treatment Group | Intervention Path | ||||||
Intervention | Control |
|
PA | Fruits/vegs | Fats/carbs |
|
|
351 (45.0) | 436 (55.0) | — | 195 (55.6) | 57 (16.2) | 99 (28.2) | — | |
44.8 (10.0) | 43.5 (11.0) | .09 | 45.3 (0.71) | 42.7 (1.32) | 44.9 (1.00) | .22 | |
|
.09 | .05 | |||||
< 35 | 63 (18.0) | 106 (24.3) | 32 (16.4) | 10 (17.5) | 21 (21.2) | ||
35-50 | 173 (49.3) | 195 (44.7) | 94 (48.2) | 37 (64.9) | 42 (42.4) | ||
> 50 | 115 (32.8) | 135 (31.0) | 69 (35.4) | 10 (17.5) | 36 (36.4) | ||
|
.42 | .16 | |||||
Women | 256 (72.9) | 329 (75.5) | 148 (75.9) | 43 (75.4) | 65 (65.7) | ||
Men | 95 (27.1) | 107 (24.5) | 47 (24.1) | 14 (24.6) | 34 (34.3) | ||
|
.005 | .80 | |||||
African American | 25 (7.1) | 33 (7.6) | 12 (6.2) | 4 (7.0) | 9 (9.1) | ||
Asian | 28 (8.0) | 39 (8.9) | 19 (9.7) | 3 (5.3) | 6 (6.1) | ||
Latino | 14 (4.0) | 18 (4.3) | 6 (3.1) | 3 (5.3) | 5 (5.1) | ||
White | 111 (31.6) | 188 (43.1) | 62 (31.8) | 21 (36.8) | 28 (28.3) | ||
Mixed/Unknown | 173 (49.3) | 158 (47.7) | 96 (49.2) | 26 (45.6) | 51 (51.5) | ||
|
.43 | .29 | |||||
High school or |
97 (27.6) | 138 (31.7) | 61 (31.3) | 12 (21.1) | 24 (24.2) | ||
College grad | 119 (33.9) | 145 (33.3) | 59 (30.3) | 20 (35.1) | 40 (40.4) | ||
Graduate/ |
135 (38.5) | 153 (35.1) | 75 (38.5) | 25 (43.9) | 35 (35.4) | ||
|
.85 | .78 | |||||
Yes | 153 (43.6) | 193 (44.3) | 88 (45.1) | 23 (40.4) | 42 (42.4) | ||
No | 198 (56.4) | 243 (55.7) | 107 (54.9) | 34 (59.7) | 57 (57.6) | ||
28.5 (6.8) | 28.7 (7.5) | .74 | 30.0 (0.47) | 25.7 (0.87) | 27.3 (0.66) | < .001 | |
|
.30 | < .001 | |||||
< 25 | 123 (35.0) | 165 (37.8) | 56 (28.7) | 29 (50.9) | 38 (38.4) | ||
25-29.9 | 117 (33.3) | 123 (28.2) | 59 (30.3) | 21 (36.8) | 37 (37.4) | ||
30-34.9 | 55 (15.7) | 63 (14.5) | 35 (18.0) | 5 (8.8) | 15 (15.2) | ||
35 and above | 56 (16.0) | 85 (19.5) | 45 (23.1) | 2 (3.5) | 9 (9.1) |
a
b
At baseline, the mean and standard deviation (SD) was 49.9 (7.9) and 48.0 (9.6) for the SF-8 Physical and SF-8 Mental summary scores respectively. The effect of treatment was significant for the two summary variables: change in these factors was significantly greater in the intervention group compared to the control group (
Effect of Alive! on SF-8 summary measures and self-assessed health status: change in the intervention group vs change in the control group
Adjusted Mean Change (MC)a or Odds Ratio (OR)b |
|
||
Variable | Intervention | Control | |
SF-8 Physicala | MC 1.84 (0.96 -2.72) | MC 0.72 (-0.15 - +1.58) | .02 |
SF-8 Mentala | MC 0.69 (-0.28 - +1.67) | MC -0.29 (-1.22 - +0.65) | .02 |
Self-Assessed Health Status (SF8 “General health”)b | OR 1.57 (1.21 - 2.04) | < .001 |
aAdjusted mean change and significance from mixed models with department as random effect factor and adjusted for baseline value, age, sex, and ethnicity.
bOdds ratio and significance, odds of having a reported improvement in general health, in the intervention group vs the control group. Model from ordinal logistic regression, with randomization group as primary fixed effect, department as random effect factor and adjusted for baseline value, age, sex, and ethnicity.
The proportion of the sample reporting greater than zero hours for difficulty concentrating and accomplishing work tasks because of back pain or depression/anxiety at baseline was 22.5% (177/787) and 30.6% (241/787) respectively (data not shown). Decrease in number of hours of back pain and depression in the intervention group vs the control group approached significance, while differences in the third presenteeism measure were significant (
Effect of Alive! on presenteeisma: change in the intervention group vs change in the control group
Variable | Odds Ratio |
|
Decreased hours of back pain at workb | 1.66 (0.99 - 2.79) | .054 |
Decreased hours of depression at workb | 1.74 (0.98 - 3.10) | .06 |
Change in Concentrate/accomplishc | 1.47 (1.05 - 2.05) | .02 |
aPresenteeism refers to the situation in which the employee is present at work, but productivity is reduced as a result of physical or mental conditions. Intention to treat models, everyone included, non-responders set to zero change. Models from dichotomous or ordinal logistic regression with department as random effect factor and adjusted for baseline value, age, sex, and ethnicity.
bOdds ratio and significance, odds of having a decrease in pain or depression, in the intervention group vs the control group. Questions were asked in following format: “During a typical 8-hour workday, about how many hours does BACK PAIN interfere with concentrating on work and accomplishing work tasks?”. Range of responses was 0-8. Change scored as 1 = hours decreased, 0 = hours stayed the same or increased.
cOdds of having improvement, intervention group vs. control group. Ordinal logistic regression with department as random effect factor, and adjusted for baseline. Question was asked in following format: “During the past 4 weeks, how much difficulty did you have concentrating at work and accomplishing work tasks because of physical or emotional problems?”
In
Persons in the intervention group had significantly greater improvement in confidence in ability to change their diet than did those in the control group (
When all subjects are included, including those in Maintenance at baseline and thus with no room to improve, there was significant or almost significant forward movement in Stage in the intervention group in comparison with change in the control group for all domains except for change in sugar (
Effect of Alive! on self-efficacy and stage of readiness for change: change in the intervention group vs change in the control group
Intention-to-treat | Intention-to-treat | |||
At-risk subjectsa | All subjectsa | |||
Odds Ratio (95% CI)b |
|
Odds Ratio (95% CI)b |
|
|
|
||||
Self-efficacy to change diet | 2.68 (1.57 - 4.57) | < .001 | 2.05 (1.44 - 2.93) | < .001 |
Self-efficacy to change physical activity | 1.42 (0.98 - 2.07) | .07 | 1.21 (0.87 - 1.67) | .26 |
Stage: Changing fat | 1.32 (1.00 - 1.76) | .05 | 1.27 (0.99 - 1.63) | .06 |
Stage: Changing fruits/vegetables | 1.76 (1.31 - 2.36) | < .001 | 1.62 (1.23 - 2.13) | .006 |
Stage: Changing added sugars | 1.84 (1.31 - 2.58) | < .001 | 1.23 (0.92 - 1.64) | .17 |
Stage: Changing physical activity | 1.42 (1.06 - 1.90) | .02 | 1.34 (1.00 - 1.80) | .05 |
aIn intention-to-treat models, subjects who did not respond to the follow-up questionnaire have their change score set to zero. CI: 95% Confidence Interval. “All Subjects”: Subjects in Maintenance (for Stage analysis) or “Very confident” (for Self-efficacy analysis) at baseline are included. “At-risk Subjects”: Excludes those in Maintenance (or “Very confident”) at baseline.
bOdds ratio: Odds of having forward movement, intervention group vs control group.
cSignificance of odds ratio for forward movement for intervention group vs control group from ordinal logistic regression models with department as random effect factor, adjusted for baseline value, age, sex, and ethnicity.
The personalized report on their diet and physical activity behaviors, which was provided to all 787 participants prior to randomization immediately after completion of the baseline questionnaires, appears to have benefited those subsequently randomized to the control group as well as those randomized to the intervention group. Of control group respondents to the follow-up questionnaires at the end of the 4-month period, 89.1% (271/304) reported they learned “Some” or “A lot” about their physical activity behaviors, and 88.5% (269/304) reported they had learned “Some” or “A lot” about their dietary behaviors (data not shown). Results were similar for the intervention group. Among members of the intervention group, 154 of 224 respondents to the follow-up questionnaires (68.8%) found the tailored tips “Somewhat” or “Very” relevant and helpful. The chat room was infrequently used. However, participation in the key element of the
The effects on SF-8 measures and self-reported general health suggest a potentially important beneficial effect of participation in the
A recent large study demonstrated that health-related productivity losses cost employers more than four times as much as medical and pharmacy costs [
The improvements in self-efficacy shown here may have important implications for the longer-term impact of participation in
We believe that the demonstrated success of
The approach of
In this study, the exact rate of participation in the randomized trial is not known, as there was no way to know how many of the 9733 email addresses were live nor how many of the invitational messages may have been spam-filtered. Our estimate is a participation rate of approximately 10%. This participation rate in the trial is reasonably consistent with other randomized trial experiences. As noted above, substantially more than 787 completed the assessments and received the feedback but did not choose to participate in the randomized trial. It is notable that there was no monetary incentive, and potential participants were told that they might not receive the intervention for 8 months if they were randomized to the control group. In addition, the participation rate was considerably higher than has been seen in some other Internet-based interventions. For example, Glasgow et al [
Engagement in this intervention was substantial, with an average of 10.9 goals selected per person over the 12 intervention sessions, and with 74% of intervention group subjects interacting with the program on 7 or more of the 12 intervention sessions. This appears to be a substantially higher engagement than some researchers have seen in Internet-based programs for the general population. Glasgow et al [
Some limitations of
It is also acknowledged that effect sizes are small in this intention-to-treat analysis in which those with missing data are assigned change scores of zero. It is worth noting that the trial randomized subjects even if they had already met diet or physical activity goals or were already at the top of scales such as efficacy and stage. Thus, the study differs fundamentally from classic “clinical” trials in which only at-risk subjects are randomized. It is also worth noting that participants chose a goal only after being randomized to the intervention or control groups, and thus each person in the intervention group participated in only one of the three intervention topics. Consequently, the generalized effects on efficacy, stage, and quality of life suggest a generalized halo effect on healthy behaviors and characteristics beyond the direct topic in which they participated.
Another limitation is the fact that there are no objective measures of outcomes like self-efficacy, quality of life, sick days, or productivity. Potential conflict of interest of some of the authors may also be noted as a limitation, as NutritionQuest developed
A notable strength of
In summary, these results show that participation in
The study was funded by a grant from the Centers of Disease Control and Prevention, Health Protection Research Initiative, grant #R01 DP000095-03. Trial Registration Number: NCT00607009. We acknowledge the contribution of Melissa Nelson of the Kaiser Permanente Division of Research, who participated in the coordination of the study; the KP staff who facilitated the implementation of the study; and all of the study participants.
GB, CB, and TB are co-owners of NutritionQuest, which holds the copyright on
Examples of
Examples of
a lifestyle intervention via email
cross-cultural activity participation study
Centers for Disease Control and Prevention
fats and sugars
fruits and vegetables
health maintenance organization
health risk assessments
Kaiser Permanente of Northern California
National Health and Nutrition Examination Survey
physical activity
worksite internet nutrition