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.
Self-monitoring is a key behavior change mechanism associated with sustained health behavior change. Although Web-based interventions can offer user-friendly approaches for self-monitoring, engagement with these tools is suboptimal. Increased use could encourage, promote, and sustain behavior change.
To determine whether email prompts or email plus telephone prompts increase self-monitoring of behaviors on a website created for a multiple cancer risk reduction program.
We recruited and enrolled participants (N = 100) in a Web-based intervention during a primary care well visit at an urban primary care health center. The frequency of daily self-monitoring was tracked on the study website. Participants who tracked at least one behavior 3 or more times during week 1 were classified as meeting the tracking threshold and were assigned to the observation-only group (OO, n = 14). This group was followed but did not receive prompts. Participants who did not meet the threshold during week 1 were randomly assigned to one of 2 prompting conditions: automated assistance (AA, n = 36) or automated assistance + calls (AAC, n = 50). During prompting periods (weeks 2–3), participants in the AA and AAC conditions received daily automated emails that encouraged tracking and two tailored self-monitoring reports (end of week 2, end of week 3) that provided feedback on tracking frequency. Individuals in the AAC condition also received two technical assistance calls from trained study staff. Frequency of self-monitoring was tracked from week 2 through week 17.
Self-monitoring rates increased in both intervention conditions during prompting and declined when prompting ceased. Over the 16 weeks of observation, there was a significant between-group difference in the percentage who met the self-monitoring threshold each week, with better maintenance in the AAC than in the AA condition (
Prompting can increase self-monitoring rates. The decrease in self-monitoring after the promoting period suggests that additional reminder prompts would be useful. The use of technical assistance calls appeared to have a greater effect in promoting self-monitoring at a therapeutic threshold than email reminders and the tailored self-monitoring reports alone.
ClinicalTrials.gov NCT01415492; http://clinicaltrials.gov/ct2/show/NCT01415492 (Archived by WebCite at
Self-monitoring of behaviors and health measures such as diet, smoking, physical activity, and weight is a key behavior change mechanism. Adherence to self-monitoring regimens is associated with greater behavior change [
Participants in Web-based interventions who self-monitored their weight or physical activity, or both, have been shown to have greater success than those who do not [
Despite its importance, there is limited information about the frequency of self-monitoring in Web-based interventions. This may be because process data are seldom reported [
The prompting study (hereafter referred to as substudy) described in this paper was a substudy of Healthy Directions 2, a randomized controlled trial of a multiple risk factor cancer prevention intervention conducted in two urban primary care health centers located in metropolitan Boston, USA. In Healthy Directions 2, patients from 33 participating providers were recruited and randomly assigned to study arms at the provider level. Enrollment eligibility included being a health center patient, being 18+ years of age, having a scheduled well visit or chronic disease management appointment, and being able to read English. Patients were ineligible if they had undergone cancer treatment in the previous year or had a diagnosis of dementia, blindness, neurodegenerative disease, or psychiatric illness (including substance abuse, psychosis, or schizophrenia in the previous 5 years).
The Healthy Directions 2 intervention targeted multiple cancer risk factors and was designed to (1) promote physical activity, (2) reduce red meat intake, (3) increase fruit and vegetable consumption, (4) promote daily multivitamin use, and (5) promote smoking cessation, as applicable. Intervention components included an endorsement of behavior change by the participant’s health care provider; materials delivered through the study website or in print, based on participant preference; intervention materials for participants’ friends and family members; and links to community-based resources. The materials emphasized the importance of consistent and continued tracking of health behaviors. The website included a user-friendly section where patients could self-monitor all targeted behaviors at once. Although daily self-monitoring was encouraged, the website allowed participants to enter data for the day they logged into the website and for the 2 days prior. After entering data, participants received immediate feedback in the form of graphs and descriptive text. Participants could also view their data over time, to assess overall progress. Intervention materials were available via the Web or as a print packet. When joining the study, each participant randomly assigned to the intervention received a bottle of multivitamins, a pedometer, intervention materials or logon information for the study website, and a US $5 gift card. The Healthy Directions 2 study was approved by the institutional review board at Harvard Vanguard Medical Associates.
After completing recruitment for the parent study, we recruited an additional 100 participants for the substudy. This substudy was separate from the parent study. Eligibility requirements were the same as those for the parent study, plus the following additional criteria: (1) having an email address, (2) having the ability to access the Internet daily, and (3) being willing to receive the Healthy Directions 2 intervention via the Web only.
Recruitment for both the parent study and the substudy was the same. Eligible patients were sent an introductory letter that outlined the study and let them know that they may be approached and invited to join the study at their upcoming appointment. At check-in, study staff met the patient and verbally introduced the study, and interested individuals provided written informed consent and completed a self-administered baseline survey.
During recruitment for the substudy, eligible participants were made aware that they may receive an additional intervention that would include emails and possibly two technical assistance calls. After completing the survey, each substudy participant received a bottle of multivitamins, a pedometer, login information for the study website, and a US $5 gift card for completing the survey. Recruitment for the substudy was limited to one site (8 providers) and took place in March 2010. As
Flow of participants through the substudy.
We set a minimum threshold of self-monitoring at least one behavior 3 or more times per week [
Participants assigned to the AA condition (n = 36) received 2 weeks of daily emails during the prompting period (weeks 2 and 3) that encouraged them to track their behaviors via the study website. Email messages changed daily and included a brief message about the benefits of self-monitoring and a hyperlink to the study website. Participants could choose to respond directly to the email with their tracking information instead of logging into the website; study staff uploaded emailed self-monitoring data to the study website. Participants also received two tailored self-monitoring reports: the first at the end of week 2 and the second at the end of week 3. Reports provided feedback to the individual about his or her frequency of tracking for each of the behaviors during the previous week. The reports mirrored the self-monitoring graphs available on the website; we hoped that seeing the graphs would encourage individuals who had not self-monitored to visit the website and track their behaviors. Reports were viewed as part of the prompting intervention. If participants did not self-monitor, their reports reiterated the information on the benefits of self-monitoring that was included on the daily emails and encouraged self-monitoring via the study website.
Participants randomly assigned to the AAC condition (n = 50) were sent the emails and tailored self-monitoring reports, detailed above, and received two technical assistance calls. The first call was made at the end of the first week of prompting (week 2) and the second call took place at the end of the second week of prompting (week 3). The calls, conducted by a trained health coach, were designed to be brief (less than 5 minutes) and focused on troubleshooting technical questions (eg, trouble logging in or how to self-monitor on the website).
We created two categories of tracking measures. The first, self-monitoring measures, focused on the frequency of monitoring. We created the second, threshold measures, to examine the minimum weekly therapeutic threshold of self-monitoring 3 or more times per week. We used multiple measures to attempt to fully capture participants’ interaction with monitoring.
We determined the total number of self-monitoring events (range 0–112). We then determined the total number of weeks during which participants self-monitored at least once (range 0–16), the greatest number of continuous weeks in which participants self-monitored (range 0–16), and the frequency with which participants self-monitored each week (range 0–7).
Using the frequency of self-monitoring each week, we determined whether participants met the weekly threshold each week (yes or no), total number of weeks during which participants met the threshold (range 0–16), and the greatest number of continuous weeks in which participants met the threshold (range 0–16).
The baseline survey included items to assess race, Latino ethnicity, marital status, frequency of Internet use, comfort level in using computers, and participants’ financial situation by asking participants to rate the “money situation” in their household (comfortable with extras, enough but no extras, have to cut back, and can’t make ends meet). Age, sex, and primary care provider was determined by data obtained from participants’ electronic medical records.
All behaviors were assessed using validated measures [
We obtained descriptive statistics for key variables (using SAS version 9.1; SAS Institute, Cary, NC, USA). We used chi-square statistics and analysis of variance to assess differences in demographics and in meeting the behavioral recommendations between those who met and those who did not meet the weekly self-monitoring threshold of tracking at least one behavior 3 or more times during week 1 (OO vs AA and AAC). We used analyses of variance to assess differences between the three groups (OO, AA, and AAC) and the self-monitoring variables and the threshold variables.
To examine the impact of the prompting conditions from baseline through the prompting period, we tested a series of 2 (group: prompting conditions) × 3 (time: weeks 1–3) repeated-measures models. Separate models were used for each dependent variable (eg, self-monitored each week, met weekly threshold of tracking at least one behavior 3 or more times/week, or frequency of self-monitoring each week). We used binomial repeated-measures models for the dichotomous outcomes and general liner models for the continuous outcomes. The models were specified with a within-group factor of time and a between-group factor of prompting condition. We conducted similar analyses to examine the impact of the prompting conditions over the 16 observational weeks. We then included the OO group in the analyses and conducted repeated-measures models with 3 (group: OO, AA, and AAC) × 16 (time: weeks 2–17) and post hoc tests to explore any differences between the groups.
The sample (N = 100) was 53% (53/100) male, with a mean age of 45.6 years, and was racially and ethnically diverse, with 37% (37/100) of the sample being black and 8% (8/100) being Latino or Latina. Most participants were college graduates (70/100, 70%) and reported frequent Internet use (see
Demographic characteristics of the study sample (N = 100).
Characteristic | Data | |
|
||
Mean (SD) | 45.6 (14.9) | |
Range | 21–84 | |
|
||
White | 52 (52%) | |
Hispanic | 8 (8%) | |
Black | 37 (37%) | |
Other | 3 (3%) | |
|
||
Male | 53 (53%) | |
Female | 47 (47%) | |
|
||
High school diploma or less | 5 (5%) | |
Some college | 24 (24%) | |
College graduate or more | 70 (70%) | |
Married, n (%) | 63 (63%) | |
|
||
Fair/poor | 7 (7%) | |
Good | 44 (44%) | |
Excellent/very good | 47 (47%) | |
|
||
Comfortable, with some extras | 49 (49%) | |
Enough, but no extras | 32 (32%) | |
Have to cut back | 15 (15%) | |
Cannot make ends meet | 4 (4%) | |
|
||
once | 5 (5%) | |
2–4 times | 14 (14%) | |
5+ times | 80 (80%) | |
|
||
Very uncomfortable/uncomfortable | 7 (7%) | |
Comfortable/very comfortable | 91 (91%) |
a Due to missing values, percentages may not total 100%.
Overall, 99.92% of participants who self-monitored, regardless of prompting condition and frequency of self-monitoring, tracked all four (nonsmokers) or five (smokers) of the targeted behaviors each time that they self-monitored (1219 of 1220 times tracked). In addition, during 76.8% (205/267) of the weeks when participants self-monitored at least once, they met the weekly threshold for self-monitoring during that week (tracking 1 or more behaviors 3 or more times). In all conditions, self-monitoring rates decreased over time.
The OO group consisted of study participants who met the weekly self-monitoring threshold (tracking 1 or more behaviors 3 or more times) during week 1 and therefore did not receive prompts. Within this group, some self-monitoring occurred in 100% of the observational weeks.
Percentage of participants, by group, self-monitoring 1 or more times per week (N = 100). The prompting intervention occurred during weeks 2 and 3. AA = automated assistance group, AAC = automated assistance + calls group, OO = observation-only group.
Percentage of participants, by group, meeting the self-monitoring threshold by week (N = 100). The prompting intervention occurred during weeks 2 and 3. The weekly threshold for self-monitoring was defined as tracking 1 or more behaviors 3 or more times during a week. AA = automated assistance group, AAC = automated assistance + calls group, OO = observation-only group.
Time had a within-group effect on whether self-monitoring occurred each week (
The repeated-measures models examining whether self-monitoring occurred each week revealed a within-group effect for time (
In the AAC condition, there were more weeks when people self-monitored their behaviors at least once than in the AA condition (16/16, 100% vs 10/16, 63% weeks,
The results of the repeated-measures models and post hoc analyses examining the three self-monitoring outcomes determined that the OO group was significantly different from both of the prompting conditions on all outcomes (
Only a small percentage of participants (9/86, 10%) replied to the tracking reminder emails with their tracking information, but of those who responded, 67% (6/9) replied multiple times.
Consistent and continued self-monitoring is an instrumental strategy for initiating and maintaining behavior change [
The purpose of this study was to examine the effect of two limited prompting interventions on increasing participants’ use of self-monitoring tools available on the study website. In this study, participants who did not meet the self-monitoring threshold of tracking at least one behavior 3 or more times during the first week of the study were enrolled in one of two prompting conditions. Both prompting conditions yielded a modest but significant increase in self-monitoring each week (yes vs no), meeting the weekly threshold (yes vs no), and frequency of self-monitoring/week during the prompting period. After prompting ceased, self-monitoring rates decreased in both prompting conditions. Although the significant within-group effect remained throughout the observation period, the increased self-monitoring during the prompting period coupled with the subsequent decline suggest that reminder emails can increase prompting, but that additional subsequent email reminders may be useful to help sustain this increase. Future research is needed to examine the effectiveness for different prompting intervals [
Over the course of the study, a greater percentage of participants in AAC, the group that received technical assistance calls, met the weekly self-monitoring threshold, suggesting that brief contact, even in the context of technical assistance calls, may be beneficial for promoting and sustaining tracking. This limited contact with study staff may have increased motivation to track by removing technical barriers, or simply served as a gentle push to self-monitor. In total, 80% of participants in the AAC condition completed call 1, and 70% completed call 2; higher completion rates may have resulted in the AAC prompting intervention having a greater impact on self-monitoring. Future research is needed to explore the nature of interactions with staff that prompt higher levels of engagement (modality, frequency, content, etc), and whether a more explicit focus on enhancing motivation to track may yield greater self-monitoring.
Over the course of the study, the OO group had the highest rates of self-monitoring each week. However, even with this group’s strong start, their self-monitoring during weeks 3–5 declined precipitously, suggesting that this may be a critical time to intervene. Beginning with week 4, when prompting ended, self-monitoring rates decreased in both the AA and AAC groups. This general across-the-board attrition is similar to those seen in other studies [
It is noteworthy that participants who self-monitored chose to track multiple behaviors simultaneously 99.9% of the time. This is encouraging and suggests that tracking multiple behaviors is not burdensome. The willingness to track multiple behaviors in this study may have been due, in part, to the beta testing of the website that was conducted to inform a user-friendly design. The website was specifically structured so that all tracking could be completed on one page, thus making it easy to track all behaviors at the same time. In addition, it is striking that in 76.8% (205/267) of the weeks when participants self-monitored they met the threshold for that week, suggesting that specific strategies could be helpful for individuals who are willing to self-monitor but do not reach a weekly therapeutic threshold. The response rate to the tracking emails was low. It is possible that participants who were motivated to track simply logged in to the website and recorded their self-monitoring data themselves without assistance. It also could be that participants found it difficult to respond to the emails because of the timing of the sent messages (ie, unable to respond to the email during the workday). One Web-based study found that participants who actively responded to emails had a greater increase in fruit and vegetable intake than did participants who did not reply to emails [
Most often tracking has been limited to pencil and paper. Burke and colleagues reported that participants found self-monitoring via a personal digital assistant to be more socially acceptable than monitoring in paper logs [
Study limitations include a modest sample size; a largely well-educated sample, which may limit generalizability; and the lack of a nonprompting control group that included only participants who did not self-monitor during the first week of the study. Study strengths include having a diverse racial and ethnic sample and a design that allowed comparison with a group already motivated to record their behavior.
It would be helpful if interventions that include self-monitoring provided information about these rates or used consistent definitions of self-monitoring. But, to date, the literature has not provided either, making it difficult to compare adherence across studies [
The 2010 US Affordable Care Act emphasizes prevention and engaging patients as active participants in their health care [
Consort EHealth Checklist V1.6 [
automated assistance group
automated assistance + calls group
observation-only group
We thank Harvard Vanguard Medical Associates for the opportunity to conduct our research in their clinical sites. This work was supported by NIH grants R01 CA123228 and 1K05 CA124415. We also would like to thank the study participants and the Healthy Directions 2 staff: Louisa Bloomstein, Joanne Chin, Lindsay ElShazly, Kristie Usher
None declared.