BACKGROUND: Health-related interventions informed by behavioral theory have been shown to be more effective in changing behaviors as compared to those that are not. The Theory of Planned Behavior (TPB) has been used to successfully predict and explain a variety of health related behaviors, including sugar-sweetened beverage (SSB) intake and physical activity (PA). The TPB assumes that behavioral intentions are the most important determinant of behavior. Intentions are the function of individual's attitudes toward the behavior (these evaluations can be positive or negative), subjective norms (social standards and expectations surrounding the behavior), and perceived behavioral control (perception of the ease with which the behavior can be performed). According to literature, behavioral intentions predict 20% - 40% of the variance in health behaviors with attitudes beings the strongest predictor of diet, and perceived behavioral control being the strongest predictor of physical activity related intentions. Excessive SSB consumption and inadequate PA have been highly associated with the obesity epidemic, and related comorbidities such as cardiovascular disease and type-2 diabetes. Understanding and targeting these behaviors through application of health behavior theories, such as the TPB, is important.

PRIMARY AIMS: This research is embedded within a larger 2-arm randomized-control trial, Talking Health, which targets residents in rural southwest Virginia. Guided by the TPB, the overall goal of the Talking Health trial is to determine the effectiveness of a 6-month intervention aimed at decreasing SSB intake (SIPsmartER) compared to a matched contact control aimed at increasing PA (MoveMore). Each condition includes three classes, one teach-back call, and 11 interactive voice response (IVR) calls. The primary aims of this secondary analysis of Talking Health are to 1) determine if single-item TPB indicators are correlated with multi-item TPB scales for SSB and PA; 2) examine how baseline TPB variables predict participation in the SIPsmartER and MoveMore; 3) determine how the IVR TPB variables assessed during IVR calls predict future SSB and PA behaviors reported in a subsequent IVR call; and 4) explore how TPB variables change over the course of the teach back and 11 IVR calls.

METHODS: Eligibility requirements included being 18 years of age or older, having reliable access to a telephone, drinking 200 kilocalories of SSB per day, and having no contraindications for moderate-intensity physical activity. The present research utilizes data from the baseline health assessment, class attendance and IVR and teach back calls completion data, as well as data collected in teach-back and 11 IVR calls. Multi-item TPB constructs for both SSB and PA behaviors were assessed at baseline (measured on a 7-point Likert scale). Each IVR call assessed self-reported past week behavior (ounces of SSB or minutes of PA) and four single-item TPB constructs including behavioral intentions, perceived behavioral control, instrumental attitudes, affective attitudes, and subjective norms. Participation was measured as the number out of 15 activities completed by participants (three classes, one teach back call, and 11 IVR calls). Statistical analysis included descriptive statistics, Chi square tests, independent T-tests, Pearson's correlations, Cronbach's α, and sequential multi-step regression models. Multiple data imputations were used to account for missing data.

RESULTS: Of the 301 participants, 81% were female and 93% were Caucasian. The mean age of participants was 48.8 ± 13.5. Additionally, 32% of participants completed high school education, 55% earned < $20,000 per year, 32% had a full time or part time job, and 33% were classified as low health literate. Single-item indicators for both SSB-TPB questions (r > 0.60) and PA-TPB questions (r > 0.69) were highly correlated with their multi-item scales. Baseline TPB variables did not predict the participation rates in either SIPsmartER (F=1.763, R2=0.057, P=0.124) or MoveMore (F=0.815, R2=0.028, P=0.541) conditions. Of the nine SIPsmartER IVR regression models, eight were significant, and the SSB-TPB variables predicted about 30% of the variance in SSB behavior. Of the nine MoveMore IVR regression models, all were significant, and the PA-TPB variables predicted about 20% of the variance in SSB behavior. In both conditions, the majority of variance was explained by behavioral intentions and the addition of other TPB variables (perceived behavioral control, instrumental attitudes, affective attitudes, and subjective norms) explained substantially less variance in the behaviors. There were no notable patterns of change in TPB variables over 11 IVR calls for either SIPsmartER or MoveMore participants.

DISCUSSION: Our findings show that single-item indicators can be used as reliable measures of the TPB constructs. The TPB model did not show significant predictive value when it comes to participation in SIPsmartER or MoveMore. On the other hand, our findings show that TPB model explained about 30% (SSB) and about 20% (PA) of variance in behavior. Although significant changes in IVR TPB variables were found between the two time points in several instances for both SSB and PA behavior, there were no patterns of change over time. Based on our findings, assessing behavioral intentions as the goal behavior in each IVR call may be the most useful application of the TPB. Other TPB variables can be assessed using single-item indicators.