This study demonstrates the feasibility of changing multiple unhealthy diet and activity behaviors simultaneously, efficiently, and with minimal face-to-face contact by using mobile technology, remote coaching, and incentives. The increase fruit/vegetable and decrease sedentary leisure treatment maximized healthy lifestyle change compared with the other interventions. In addition to producing targeted improvements in fruits/vegetables and sedentary leisure time, the treatment produced untargeted improvement in saturated fat intake. The superiority of the increase fruit/vegetable and decrease sedentary leisure treatment was present after 1 week of intervention persisted through the end of the 3-week treatment phase and was maintained. As expected, since participants were no longer asked to maintain healthy changes, lifestyle gains did diminish once treatment ended. Nevertheless, substantial improvements (1 s.d. compared to baseline) in fruit/vegetables, sedentary leisure, and fat persisted through the 5-month follow-up. From baseline to end of treatment to end of follow-up, respectively, mean fruits/vegetables changed from 1.2 to 5.5 to 2.9; mean minutes/day sedentary leisure from 219.2 to 89.3 to 125.7; and percent daily calories from saturated fat from 12.0 to 9.4 to 9.9. Even though they were neither asked nor reinforced to maintain eating or activity improvements, 86% of the 185 participants from whom exit interviews were obtained said they “definitely” or “somewhat” tried to maintain gains.
Consistent with behavioral choice theory,23–24
decreasing recreational screen time was complemented by a reduction in saturated fat. Previous research has demonstrated that manipulating screen time changes energy and fat intake in children.38
This is the first study to show that reducing sedentary leisure time decreases fat intake in adults. Increased interest in sedentary behavior has been driven by epidemiological data associating sedentarism with cancer39
, type 2 diabetes 19, 40–42
, cardiovascular events19,44
, and increased mortality 19, 44–45
, independent of physical activity. Reduced screen time may be an important behavioral target not only to reverse direct adverse effects of prolonged sitting46
but also to disrupt pairing of screen time with high-fat snacking.
Notably, the traditional dieting regimen (decrease fat and increase physical activity) produced less healthy change than other treatments: it only improved two behaviors, and increased physical activity did not persist. The requirement to inhibit rewarding behaviors had no systematic impact on diet-activity change: the increase fruit/vegetables and physical activity treatment was no more advantageous than other treatments and the decrease fat and sedentary leisured was no more disadvantageous. 24
These results are germane to physicians trying to help patients improve multiple health risk behaviors. Physicians play an important role by advising and assisting patients to accomplish healthy behavioral changes,9,47
especially since a trusting relationship with a provider is associated with greater adherence to advice.3
However, limits on physicians’ time combined with movement towards new systems of patient-centered, team-based care48
create an opportunity to reconsider the optimal locus and configuration of health behavior change counseling. Results suggest feasibility and potential benefit of a systems reconfiguration that reinforces health behavior change by connecting patients with mobile technology, incentives, and remote, non-physician coaches.
A number of study limitations warrant consideration. Generalizability of the findings is limited by the constraints that the study was conducted in a research setting and only a quarter of the sample was male. Use of a screening phase to confirm the presence of the risk behaviors may additionally limit generalizability to those with entrenched unhealthy diet and activity behaviors. Also, the amount of the financial incentive was larger than would be feasible for some settings. It remains to be determined whether such rapid and full acquisition of behavior change targets would occur with smaller incentives. Further, the fact that primary outcome measures were self-reported raises the possibility that participants might have overstated their behavioral improvements to earn incentives. We find that unlikely for several reasons. First, treatment differences remained after controlling for the effects of financial motivation and social desirability. Second, the sample ranked financial motives lowest among their reasons to join the trial. Third, maintaining diet and activity improvements yielded no financial reward during follow-up, an altered contingency made apparent to participants by staff reminders and by discontinuation of study procedures (urine samples, accelerometry, grocery receipts) that could have verified self-reports. Yet participants maintained substantial improvements and most said they did so intentionally.
Finally, although physical activity was increased by treatments that targeted it, it was the one behavior not improved by the increase fruit/vegetable and decrease sedentary leisure treatment. We currently are testing whether all four risk behaviors can be improved by targeting physical activity simultaneously or sequentially with fruit/vegetables and sedentary leisure.
Strengths of the study include the ethnic diversity of the sample and minimal loss to follow-up. Also, the sample was deliberately chosen to present challenges for behavior change. The requirement that participants unremittingly display all four risk behaviors throughout baseline screening, even while self-monitoring, yielded a sample with risk behaviors that were refractory to lower intensity behavioral intervention. A key innovation was the use of mobile technologies that connect and provide decision support to patients and coaches, reducing the need for professionals to perform counseling. Another strength was the conservative, comparative research design that contrasts active treatments.
Interventions that target multiple, prevalent, covariant risk behaviors simultaneously have the potential to be powerfully efficient and cost effective. Yet many multiple behavior change interventions have achieved limited success,49–51
presumably because their interventions were insufficiently intensive.52
As mobile technologies become increasingly ubiquitous, they afford a scalable platform to extend continuing support for healthy behavior change pervasively into the environment with potential to improve population health.