Design and rationale for a randomized trial of a theory- and technology- enhanced physical activity intervention for Latinas: The Seamos Activas II study

Abstract

Introduction:

Latina women report disproportionately high rates of physical inactivity and related chronic health conditions. Physical activity (PA) efforts to date have shown modest success in this at-risk population; thus, more effective interventions are necessary to help Latinas reach national PA guidelines and reduce related health disparities. This paper describes the design, rationale, and baseline findings from the Seamos Activas II intervention.

Methods/Design:

The ongoing RCT will test the efficacy of the Seamos Saludables PA print intervention vs. a theory-and technology-enhanced version (Seamos Activas II). The purpose of the study is to increase the percentage of Latinas meeting the national PA guidelines compared to the prior trial, improve biomarkers related to disease, and extend generalizability to a broader and more representative population of Latinas (i.e. Mexican/Mexican-Americans). Intervention refinements included further targeting key constructs of Social Cognitive Theory, and incorporating interactive text message-based self-monitoring strategies. The primary outcome is change in minutes per week of MVPA measured by ActiGraph GT3X+ accelerometers at 6- and 12-months. Secondary PA outcomes assessed by the 7-Day PA Recall will be used to corroborate findings.

Results:

Participants (N=199) are Latinas 18-65 years (mean=43.8) of predominantly Mexican origin (89%). At baseline, objectively measured MVPA was 39.51 min/week (SD=71.20, median=10) and self-reported MVPA was 12.47 min/week (SD=22.54, median=0).Participants reported generally low self-efficacy and higher cognitive vs. behavioral processes of change.

Conclusion:

Addressing interactivity and accountability through text messaging, and more rigorously targeting theoretical constructs may be key to helping Latinas achieve nationally recommended PA levels and thereby reducing health disparities.

Introduction

Engaging in regular physical activity (PA) is an established mechanism for preventing type 2 diabetes, heart disease, select cancers, and for maintaining a healthy body weight [1]. Maintaining healthy PA levels is particularly important for Latinas in the U.S. as they are 70% more likely to be diagnosed with diabetes and 20% more likely to be obese than non-Latina white women [2–4]; yet only 44% of Latina women (vs. 55% of non-Latina white women) [5] meet the national PA guidelines (i.e., at least 150 minutes/week of moderate intensity, or 75 minutes/week of vigorous intensity, aerobic PA or a combination of both) [6]. The high rate of inactivity among Latinas requires urgent interventions to reduce related health disparities in this community [7–9].

The low PA in Latinas may be due to a number of social and environmental barriers, such as language, cultural norms, poor social support, lack of transportation, family and work obligations [10–13], as well as personal barriers including lack of time, fatigue, and lack of interest in exercise [13–16]. Home-based interventions help overcome many of these barriers and are likely more cost-effective and sustainable than face-to-face approaches [17–21]; however, most (recent) [22–24] efforts to date have been center-based.

The few available home-based interventions have shown potential in this high-risk population. The resulting PA increases are rather modest though, rarely reaching the national guidelines [25–27]. For example, in the Seamos Activas and Seamos Saludables studies [27–29], theory-driven culturally and linguistically adapted, mail-delivered PA interventions produced significant increases in weekly minutes of moderate to vigorous PA (MVPA) from baseline to six months [29]. However, only 37.8% and 11.36% (respectively) of these intervention participants reported meeting national guidelines for PA at 6 months. Thus, to reach criteria for health-enhancing PA and maintain these gains long-term, additional enhancements to the intervention were necessary.

To address this need, the Seamos Saludables [28–30] program, a PA intervention emphasizing key constructs of Transtheoretical Model (TTM) [31] and Social Cognitive Theory (SCT) [32] (i.e., processes of change, self-efficacy, self-regulation), was refined to achieve greater PA levels. These refinements included: 1) modifying print-based materials to target constructs of SCT [32] that were not directly addressed in the original intervention but increased among participants with the highest PA increases (e.g., enjoyment/outcome expectations and social support), and 2) incorporating the use of interactive text-messaging in response to participant feedback for greater interactivity and accountability and increasing trends in technology use (e.g., 96% of Latinos now own a cellphone) [33–35]. Furthermore, the previous study was tested among Latinas in New England, who were mostly (75%) Dominican, Colombian, and Puerto Rican. While Mexicans and Mexican-Americans represent the largest Latino subgroup in the U.S. [36], they were underrepresented in this study (i.e., comprised 5% of participants). Thus, Seamos Activas II focused on recruiting primarily Mexican and Mexican-American Latinas in southern California to improve the program’s generalizability to a broader and more representative population.

The current study aims to test the efficacy of the theory- and technology-enhanced intervention for increasing MVPA compared to the previously tested print-only Seamos Saludables [28–30] PA intervention. The purpose of the study is to increase the percentage of Latinas meeting the national PA guidelines and improve biomarkers (HbA1c, triglycerides, LDL, HDL) related to chronic disease risk. The current study will fill knowledge gaps by addressing the paucity of technology-supported home-based PA interventions for Latinas and expand the literature by shifting the focus to achieving the long term, health enhancing levels of physical activity needed to resolve related health disparities in this community. This paper describes the design, rationale, and baseline findings of Seamos Activas II.

Methods

Design

Seamos Activas II is a 6-month randomized controlled trial comparing two PA intervention conditions in Latina women: (1) the original Seamos Saludables [28–30]empirically supported, culturally and motivationally tailored print-based PA intervention (Original Intervention), vs (2) a theory- and technology-enhanced version of the Seamos Saludables intervention (Enhanced Intervention). The primary outcome is change in minutes-per-week of moderate-to-vigorous PA (MVPA) from baseline to six months, assessed using ActiGraph GT3X+ accelerometers, with maintenance of behavior change assessed at 12 months. Secondary outcomes include change in self-reported MVPA assessed by the 7-Day Physical Activity Recall [37] (from baseline to six months and to 12 months), changes in psychosocial and theoretical mediators of PA change (from baseline to six months and 12 months), and changes in cardiovascular and metabolic biomarkers (from baseline to six-months only).

Setting and Sample

The study was conducted at the University of California San Diego (UCSD). The sample comprised women between the ages of 18-65 who met the following inclusion criteria: self-identifying as Mexican, Mexican-American, or Latina/Hispanic of any country of origin; being underactive, defined by self-reporting less than 60 minutes per week of MVPA; and owning a cell-phone with text messaging capability. Accelerometer data were not used at baseline to verify activity levels for eligibility purposes because of the use of self-report data provides a more immediate insight into participants’ activity levels, which allowed us to make eligibility decisions immediately and to proceed with timely enrollment activities. Exclusion criteria included any health condition that would make PA unsafe as reported on the Physical Activity Readiness Questionnaire (PARQ) [38] screening tool, taking any medication that may impair PA tolerance or performance (e.g., beta blockers), BMI over 45 kg/m², history of a psychiatric hospitalization in the previous 3 years, planned or current pregnancy, or plans to move out of the area within the 12-month study period. Participants who did not meet eligibility requirements based on their responses to the PARQ or other health-related screening criteria were allowed to seek medical clearance from their doctor or healthcare provider in order to participate in the study. All study activities have been approved by the University’s Institutional Review Board.

Recruitment and Eligibility Screening

Participants were recruited using methods that had proven to be successful in our previous studies [26, 29] (e.g., posting and distributing flyers, recruiting at health fairs and community events, ads in local Spanish-language newspapers and radio, Craigslist ads), as well as a Facebook page for Seamos Activas II that was developed specifically for recruitment in the current study. Latinas who were interested in participating called the research center to complete a telephone eligibility screener. All recruitment activities took place in San Diego County, California, which is comprised of a substantial Latino population (33%) of predominantly of Mexican origin (90%) [39]; thus, ensuring representation of Mexican/Mexican-American participants.

Protocol

Eligible women then attended an in-person orientation session to receive further information about the study. During this orientation session, participants’ height, weight, and self-reported PA were assessed to confirm eligibility. Participants also completed the Short Test of Functional Health Literacy in Adults (STOHFLA) [40, 41] which tests literacy for English and Spanish speaking patients in healthcare settings. This scale was used to determine study eligibility such that participants scoring below 16 (out of 36) were deemed to have an inadequate level of functional health literacy to participate in the study. Participants who were eligible and interested in participating completed the informed consent process. During a second visit to the study site, participants had their measurements taken (hip and waist circumference, percent body fat assessed by body impedance measurement, and blood pressure) and were given an ActiGraph GT3X+ accelerometer to wear for the next seven days during waking hours.

Baseline/Randomization Assessment

One week after the measurement session, participants return for a baseline/randomization assessment visit. They return their accelerometer, which is checked for proper wear-time using well-vetted protocols from previous trials [26, 29]. Accelerometer wear is considered sufficient if participants wore it for either: a) at least 5 days, with a total of ≥ 600 minutes per day; or b) 3000 minutes of wear time across 4 days. Participants then perform a supervised 10-minute treadmill walk (3-4 miles per hour) meant to demonstrate moderate-intensity level PA, followed by completion of the 7-Day PAR interview. Participants complete blood draws and questionnaires, then are randomized to one of the two PA conditions (described in detail below): 1) Original Intervention arm, or 2) Enhanced Intervention arm. Randomization was stratified by PA stage of change [42] to ensure equal distribution of the different levels of motivational readiness for PA between groups

Six- and Twelve-Month Assessments

Prior to the six- and twelve-month assessments, participants are mailed an ActiGraph GT3X+ accelerometer with detailed instructions to wear for seven continuous days before each study visit. At the six- and twelve-month assessment visits, participants return their accelerometer and a research assistant verifies sufficient wear time using the same protocol as baseline visits. Participants complete the same assessments from baseline (i.e. anthropometry, blood pressure, blood draws at 6-month only, and psychosocial measures), a treadmill walk, and the 7-day PAR questionnaire. Participants receive compensation for their time, travel and childcare to attend all study visits. Participants also complete questionnaires on program satisfaction at 12 months. Six- and twelve-month outcomes will be reported in the future.

Measures

Demographics

At baseline, participants completed demographic questionnaires to provide information regarding their age, education, income, race, marital status, occupation, country of birth and length of residence in the U.S. The Brief Acculturation Scale assesses language use across four contexts. This measure has demonstrated good internal consistency in previous research (Cronbach alpha=.90) and has shown validity per correlations with generational status, time in country, and ethnic identity [43].

Physical Activity

The primary PA outcome, objectively-measured PA is assessed using Actigraph GT3X+ accelerometers, which measure movement and intensity of activity and have been validated against heart rate telemetry [44] and total energy expenditure [45]. Participants were asked to wear the ActiGraph GT3X+ accelerometer on their left hip during seven days for a minimum of 12 hours per day. Participants were asked to re-wear the accelerometer if they did not meet a minimum wear-time requirement of: a) at least 5 days, with a total of ≥ 600 minutes per day; or b) 3000 minutes of wear time across 4 days. Data were processed using the ActiLife software, 60 second epochs, and with a cut point of 1952 for moderate PA [46]. Only bouts of at least 10 minutes of MVPA were considered, per 2008 National Physical Activity Guidelines for Americans [47].

Self-reported PA was assessed at baseline, six- and twelve-months via the 7-Day Physical Activity Recall (7-Day PAR), a semi-structured interview that provides an estimate of total weekly minutes of PA (leisure time, occupational and transport). The 7-Day PAR has been used to assess PA in many studies and has consistently demonstrated acceptable reliability, internal consistency, and congruent validity with other more objective PA measures [48–50], including Latino participants [51]. The current study uses both objective and self-reported PA measures as the original parent trial (Seamos Saludables) [28–30] was powered on the 7-Day PAR measure; thus, allowing for comparison of results to the prior study. Moreover, use of the 7-Day PAR provides contextual PA data (e.g., types of activities Latinas are engaging in) and capturing PA that may not be measured via accelerometry (e.g., swimming).

Psychosocial Measures

Participants complete the following measures monthly as well as at the baseline, 6- and 12-month assessments. All baseline, six- and twelve-month assessments are in-person, while monthly assessments are conducted by mail or online. The Stages of Change for Physical Activity [42] measure has successfully been used to stage-match treatment in different studies [42, 52, 53], and has shown acceptable test-retest reliability (Kappa = 0.78) [42] and concurrent validity with measures of self-efficacy and self-reported PA levels [42, 53]. The Processes of Change for Physical Activity (POC) [52] measure is a 40-item questionnaire that contains 10 subscales that measure the cognitive (increasing knowledge, awareness of risks, consequences to others, comprehending benefits, increasing healthy opportunities) and behavioral processes of change (substituting alternatives, enlisting social support, rewarding oneself, committing oneself, reminders). Items on the questionnaire include Likert-type response options ranging from 1-5 to indicate how often each process is used (1= never/0 days weekly, 5=always/7 days weekly). Each subscale is composed of a mean of 4 items, which are then used in the into final two subscales (Cognitive Processes and Behavioral Processes) with higher scores indicating greater use of each Process. Internal consistency of the subscales in previous research has ranged from .62 to .89 [52]. Data from the current study show a similar pattern of internal consistency. The Self-Efficacy for Physical Activity (SE) [42] is a 5-item measure used to assess self-efficacy to become physically active across diverse contexts. Response options range from 1-5 (1= not confident to 5= extremely confident) to indicate confidence in the individual’s ability to exercise in various contexts (e.g., when tired, on vacation, when lacking time). The measure is scored by calculating a mean for all 5 items, with higher scores indicative of greater self-efficacy. This scale has shown acceptable test-retest reliability (.90) and internal consistency (e.g. alpha = .82, alpha =.76) in previous research [42], and alpha=.85 in the current study. In addition, previous studies have shown concurrent and predictive validity among different populations including Latina women [54–56]. In this study, SOC, POC and SE scores serve as an input into the expert system, generating tailored reports; additionally, POC and SE scores will be assessed as mediators of the intervention effects on PA.

The Social Support for Exercise (SSE) scale ^[57] is a 13-item measure that has 3 subscales evaluating behaviors and attitudes of family and friends toward participation in exercise evaluating behaviors and attitudes of family and friends toward participation in exercise. Response options range from 1-5 to indicate frequency of support behavior (1= none to 5=very often). The questionnaire is scored by summing items in each subscale, with higher scores indicating higher social support for PA. The SSE has demonstrated acceptable internal consistency (alphas .61-,91) and criterion validity.[57] The Physical Activity Enjoyment Scale (PACES) [58] is a 18-item questionnaire assessing level of personal satisfaction derived from engagement in PA that uses a bipolar Likert scale with response options ranging from 1-7 (1 = satisfaction/enjoyment, 7 = dissatisfaction/dislike). Higher scores on this measure indicate greater satisfaction with PA. The measure has shown high internal consistency (alpha = 0.93) and test-retest reliability in our current study as well as previous research [58].

Participants are also asked to complete the Perceived Stress Scale (PSS) and the Center for Epidemiological Studies Depression Scale (CES-D) in-person during their baseline, six- and twelve-month study visits. The Perceived Stress Scale (PSS) [59] is a 14-item questionnaire that examines the degree to which specific situations are deemed stressful during the previous week. Likert type response options ranging from 0-4 are used to indicate frequency of stressful events in past month (1=never, 5=very often). The measure is scored by reverse scoring positive items, then creating a sum score for all items, with a higher score indicative of higher perceived stress. The PSS has shown good reliability and validity among different populations, including Latino populations [60–62]. The 10-item Center for Epidemiological Studies Depression Scale (CES-D) [63] assesses depressive symptoms and has shown acceptable reliability and validity in previous studies with Latino populations in the United States [64, 65]. Response options range from 0-3 to indicate how often participants experienced various depression symptoms during the past week (0= rarely to 3= most days). Higher scores on the measure are indicative of higher depressive symptoms. Positive items are reverse scored, then a sum score is created for the measure, with a higher score indicating greater depressive symptoms. At the 12-month assessments, participants complete a Consumer Satisfaction Measure that has been used by our research team in past trials [21, 66–69] and adapted to assess the feasibility/acceptability of the enhanced PA intervention among Latina women in the current study.

Biomarkers

Fasting blood samples are collected at baseline and six months to assess hemoglobin A1C, lipids (triglycerides), high-density lipoprotein cholesterol (HDL), low density lipoprotein cholesterol (LDL), and inflammation (through high-sensitivity C-reactive protein (HS-CRP). Samples (12 ml of fasting blood) are collected and stored at the University’s NIH-funded Clinical and Translational Research Institute.

Anthropometric Measures

Sitting blood pressure, body weight, and body composition (percent body fat) are assessed at baseline, six and twelve months. A mercury manometer is used to obtain sitting blood pressure, and a Health-O-meter medical scale is used to measure body weight (to the quarter pound) and height (to the quarter inch). The Quantum II bioelectrical body composition analyzer (RJL Systems, Inc., Detroit, MI) is used to estimate percent body fat from body impedance measurements.

Original Intervention Arm

The existing empirically-supported, individually tailored Spanish-language intervention from the Seamos Saludables study [28–30] is based on SCT [32] and TTM [31] and addresses key constructs of these theories (processes of change, self-efficacy, self-regulation). The original Seamos Saludables intervention was previously culturally and linguistically for this population through an extensive formative research process with Larinas (i.e., 25 cognitive interviews and 6 focus groups) to ensure clarity of intervention messages and address culture-specific attitudes and barriers to PA for Larinas [27]; thus, no further modifications were made in the Original or Enhanced intervention for Mexican and Mexican-American women in the current study. The intervention emphasizes behavioral strategies for increasing PA levels, including goal setting, self-monitoring, increasing social support for PA, and self-efficacy. Participants in the Original Intervention arm receive mailings weekly in month one, twice weekly in months two and three, and monthly during months four to six. Two booster mailings are sent during the maintenance phase, at months eight and ten. Mailings consist of: 1) motivation-matched manuals with PA information specific to participant’s stage of change for increasing PA, 2) individually tailored feedback reports, and 3) PA tip sheets that address barriers to PA in Larinas (e.g., caregiving responsibilities, neighborhood safety) identified in our formative research with Latina women [27].

Reports are individually-tailored based on participant’s answers to monthly questionnaires (i.e., self-efficacy, stages of change [42], processes of change [52]). A computer expert system draws appropriate messaging on stages and processes of change and self-efficacy, as well as progress and normative feedback from a bank of over 330 messages that address different levels of psychosocial and environmental factors affecting PA. Additionally, participants complete a goal-setting session based on motivational interviewing at their baseline and six-month visits, and via phone call at one-month post-baseline. The goal setting session involves a discussion with a trained research staff member during which participants establish realistic PA goals that can be gradually increased until they met the study goal of ≥150 minutes per week of MVPA, which is currently the national recommendation for aerobic PA [1]. This goal-setting session included identification of potential barriers and solutions. Participants keep their written PA goals and a copy is maintained in their study file so a research staff member can discuss progress during scheduled intervention calls and follow-up assessments. Participants also receive a pedometer with instructions to wear daily during all waking hours, as well as monthly PA logs to record their daily minutes of MVPA and number of steps per day to encourage self-monitoring throughout the study. At 12 months during a brief final study session, participants are assisted to develop a plan for maintaining PA progress.

Enhanced Intervention Arm

Participants in the Enhanced Intervention arm receive all the Original Intervention arm components, using the same frequency for mail-based print materials (tip sheets and tailored reports). Results from the parent study and existing literature [20, 70, 71] supported the continued inclusion of all theoretical components of the original intervention (cognitive and behavioral processes of change, self-efficacy, self-regulation), and guided the development of additional Enhanced Intervention content. We modified our print-based materials from the Original Intervention to further target theoretical constructs of the SCT that were not directly addressed, but increased more among participants with highest vs. lowest increases in PA, in our prior Seamos Saludables [29, 30, 67] trial (i.e., enjoyment/outcome expectations and social support). These modified components (described in further detail below) include: 1) more in-depth tailored reports and tip sheets, 2) interactive text messages with PA tips and prompts to self-monitor minutes of MVPA, 3) additional phone calls at months 2, 3, and 9, and 4) an individual report mapping safe, low- or no-cost places to do PA (e.g., walking routes, parks, gyms) near their home or work places.

Tailored Reports

To further target constructs of SCT that were associated with increased PA in our prior trial, tailored reports are enhanced with feedback on participants’ individual use of SCT constructs including social support, enjoyment of activities done over the past month, and with reinforcement/ encouragement to continue doing so. Feedback on participants’ reported expectations, with strategies for decreasing negative expectancies and increasing positive expectancies, is also provided in tailored reports. Information has also been added to the tip sheets highlighting the importance of SCT constructs including social support, different types of social support for PA, and ways to increase enjoyment with suggestions of activities to try.

Interactive Text Messaging

Self-monitoring has been shown in meta-analyses to be one of the most effective components of behavior change interventions [72]. The use of text messaging allows for daily self-monitoring and real-time accountability (without reliance on Internet access or smartphone technology), features which, along with greater interactivity, were specifically requested by participants in the Seamos Saludables [29, 30, 67]. Thus, given the high cell phone ownership in Latinos [33–35], we added a text messaging component to the Enhanced Intervention arm. Participants in this group receive one text per day for the first six months. The automated text messages included PA tips addressing SCT constructs (4 per week), goal-setting (1 per week), and self-monitoring (2 per week). A sample week of text messages is listed in Table 1. The goal-setting and self-monitoring texts prompt specific responses from patients (i.e., minutes of PA they plan to do or have done that week). Automated feedback texts are then sent in response to these responses. For example:

Study: “How many minutes of physical activity did you complete this week?”

Participant: “30”

Study: “Your physical activity goal this week was 30 minutes, and you reported doing 30 minutes. Great job reaching your goal this week!”

Table 1.

Sample week of text messages sent to participants in the Enhanced Intervention arm

Day	Theoretical Construct Targeted	Example of Text Message
Monday	Goal setting	Time to set your first exercise goal! What would you like your exercise goal to be this week? Thanks for setting a goal!
Tuesday	Social support	Ask [social supporter 2*] tonight to take a brisk walk with you in the morning-it will energize you for the day. *Participants are asked at baseline to name 3 potential sources of social support for PA
Wednesday	Environment	A quick Internet search can help you find fitness related classes, events, and other information in your neighborhood.
Thursday	Self-monitoring	How many minutes of exercise have you done so far this week? Thanks for reporting your minutes of exercise!
Friday	Enjoyment	Choose active entertainment over passive entertainment.
Saturday	Social support	Make new friends! Exercise classes are great for meeting new people, which makes exercising more enjoyable.
Sunday	Self-monitoring & weekly goal feedback	How many total minutes of exercise did you do in the last 7 days? Thanks for reporting your minutes of exercise! Last week your goal was 150 minutes, and you reported 175 minutes of exercise.

Telephone Calls

To reinforce and encourage text usage, enhanced arm intervention participants receive additional phone calls from a research staff member at two- and three-months post-baseline specifically to address text message participation. During these calls, participants who are not engaging with the texting component of the program are encouraged to identify and address barriers to responding to text messages. In months 7 through 12, participants receive one text message per week asking them to set a goal and one text message per week asking them to report their total minutes for the week (but they no longer receive daily text messages with tips).

Individual Report Mapping PA Locations

At baseline, participants in the Enhanced intervention arm were provided with additional print materials (i.e., individual report mapping PA locations near participants home) to help them navigate the built environment, which recent literature suggests may be particularly relevant for Larinas [20]. Specifically, after randomization, an intervention RA provided participants in the Enhanced Intervention arm a map with places to be active tailored to locations of their choice (most often this was participants’ homes and/or workplaces). Based on the addresses participants provided, the RA used Google maps and local knowledge to generate a tailored map for participants detailing safe routes to walk and/or do other activities (e.g., exercise at a park, local gyms) within the vicinity of the addresses provided.

Intervention Fidelity

The following strategies of intervention fidelity were implemented in both intervention arms in this study. First, extensive follow-up was conducted on any returned intervention mail sent to participants. Specifically, research staff contacted participants (via phone call, email and text) to verify their correct address and receipt of study materials. Participants mailing address was also verified at each in-person visit. Second, research staff members monitored receipt of participants’ monthly psychosocial questionnaires and contacted participants when their surveys were not received or completed online. Third, participants completed a consumer satisfaction questionnaire at the end or the study that assessed the extent to which study materials were received, read, and relevant for increasing PA. Additionally, for participants in the Enhanced Intervention arm only, the interactive text messaging system monitored the frequency and content of participants responses to weekly text messages.

Sample Size Considerations

Our sample size calculation was based on the assumption that we would have 80% power for testing the null hypothesis that the intention to treat PA effect is zero, versus the two-sided alternative that the PA effect is different for participants randomized to the Enhanced condition versus those randomized to the Original condition. Effect sizes used in our power calculation were derived from the parent trial (NR011295) in which mean minutes/week of PA at 6 months was 73.37 (SD=89.73) for the Intervention arm. We hypothesized that the additional enhancements would translate into an additional 35 min/week of PA at 6 months compared to the Original arm. Based on our prior studies, average one-day activity rates were 35 minutes and thus we powered the study with the assumption that enhancements would translate to an additional one day/week of activity. With 125 participants randomized to each arm at baseline, we expected to have sufficient power (80%) to detect differences in mean PA minutes at 6 months between the Enhanced Intervention and Original Intervention conditions, using a two-tailed significance level α=0.05.

Proposed Analysis of Main Outcomes

Data analysis will focus on testing differences in PA minutes between the Enhanced Intervention condition and the Original Intervention condition at baseline, 6 and 12 months. The primary variable of interest will be the number of minutes of at least moderate intensity activity (objectively measured via accelerometer). We will use a single mixed effects regression model implemented with a random intercept, in order to test both the primary question of interest (differences between arms at 6 months) and the additional aim (testing differences in PA during maintenance phase months 6-12). Random intercepts allow for adjustment of standard errors for repeated measures of the outcome over time. Models will adjust for any variables not balanced by randomization. Analysis will be conducted on the Intent to Treat sample with all participants randomized included in the analysis. Mixed effects models use a likelihood-based approach to estimation and thus do not require any direct imputation of missing outcomes. To avoid the effects of potential outliers, we will apply a normalizing transformation to the response measure (minutes of reported activity at follow-up) before proceeding with the analysis. A similar series of regression models will be used to assess between-group differences in secondary measures (self-reported MVPA, biomarkers).

As a secondary step, we will compare the efficacy of the print-based intervention arm to the results in the original Seamos Saludables study. That is, after adjusting for study differences (including potential confounders such as acculturation, country of origin), we will compare mean minutes of PA/week between the print-only condition (in the current study) and the Intervention arm (from the parent study) using a mixed effects model, similar to that described above.

Potential Moderators

A variable will be considered a moderator if evidence exists of either qualitative or quantitative interaction with the intervention. The dependent variable (i.e., minutes of at least moderate intensity PA at follow-up) will be regressed on the moderator (i.e., acculturation), independent variable (i.e., treatment assignment), and the product of the moderator and independent variable. Evidence of moderation exists if the coefficient of the interaction term is statistically different than zero. We will use a series of mixed effects models to test the potential moderators, similar to that described for the primary aim. Potential moderators include: demographics, acculturation, and PA stages of change.

Potential Mediators

Our primary approach to establishing mediation effects will be a multiple mediation approach, in which all potential mediators (SCT and TTM psychosocial constructs targeted by the intervention, such as processes of change, self-efficacy, social support, enjoyment) are tested simultaneously, using a product of coefficients method with bootstrapped standard errors (10,000 samples). The effect of the intervention on change in the mediators (a path), the effects of the change in the mediators on the change in PA at follow-up (b path), the total effect (c path) of the intervention on PA at follow-up and the indirect effect (ab) of the intervention on outcome will all be examined. Following the guidelines presented in Preacher and Hayes, [73] a construct was considered a mediator of the intervention effect if the indirect effect of treatment through that construct was significantly different than zero.

Missing Data Approaches

In the event of missing data, we will contact participants immediately. If a participant drops out of our study, we will attempt to gather follow-up information. If participants refuse to be contacted or otherwise lose contact with the investigators, we will censor the data at the point of drop-out. We will apply two statistical approaches to handling missing data and compare these to the effect estimates from the primary outcomes analysis (which take a likelihood based approach to estimation but do not directly impute missing data). The first approach is inverse probability weighting with propensity scores. [74] This is a two-step method: first we will model the probability of missingness as a function of baseline covariates and previous outcomes (using logistic regression). The resulting predicted probabilities of dropout will serve as the propensity scores. Next, the inverse propensity scores serve as weights in our regression model of the primary outcomes (i.e., reported weekly minutes of PA). Provided the data is missing at random (MAR) or that the probability of missingness can be fully explained by observable data, this approach produces asymptotically unbiased estimates. To allow for the possibility that the MAR assumption may not hold, we also will use pattern mixture models. In this case, the distribution of the primary outcome is assumed to follow a mixture of two distributions: one for those who complete follow up and another for those who do not. This method allows us to quantify the robustness of the study findings to a range of missing data assumptions [75].

Results

A total of 1838 Latinas expressed interest in participating in the study, of which 998 were screened for eligibility. Of these, 784 were excluded (e.g., not meeting inclusion criteria, declined to participate, unable to contact), with 214 participants randomized and a final analytic sample of 199 (15 participants were ineligible post-randomization). A full consort diagram is presented in Figure 1. Participants (N=199) randomized to one of the two treatment conditions at baseline (N=102 Enhanced vs N=97 Original Intervention). As shown in Table 1, the sample was predominantly Mexican American (89%), with 41% reporting at least some college level education. The average age of participants was 43.8 years (SD=10.11) and mean BMI at baseline was 30.6 (SD=7.56), with the majority (86%) of participants being overweight/obese. Mean percent body fat was 39.53 (SD=6.88), and mean blood pressure was 111.52 (SD= 13.92) over 70.47 (SD=9.07). The only between-group differences in demographic and psychosocial variables at baseline were in income- with a greater proportion of low income participants in the Original Intervention arm (p<.05), and enjoyment (p=.04)-with participants in the Enhanced intervention arm reporting higher mean scores vs. the Original intervention arm, 89.74, SD=20.14 vs. 83.95, SD=19.89, respectively. The majority of participants were in Precontemplation or Contemplation at baseline, 91.2% of Enhanced participants and 91.7% of Original Intervention participants, with no between-group differences (p>.05).

Figure 1.

Consort Diagram

Objectively measured MVPA at baseline was 33.61 min/week (SD=71.29) among Enhanced participants vs. 45.72 min/week (SD=70.93) among Original Intervention participants. Self-reported MVPA at baseline was 14.3 min/week (SD=25.06) among Enhanced group participants (median=0) compared to 10.5 min/week (SD=19.47) among Original Intervention condition participants (median=10.0). Blood samples are currently being processed and will be reported in forthcoming publications.

In terms of psychosocial variables, self-efficacy scores averaged approximately 2.0 indicating relatively low self-efficacy for exercising when tired, on vacation, lacking time, etc. Processes of change scores for cognitive processes (2.6) was higher than those for behavioral processes (2.0) but generally low overall. Baseline social support scores showed that the mean friend social support score 14.6 (SD=6.3) was lower than mean family social support score 16.7 (SD=7.4). A full description of the study sample is presented in Table 2.

Table 2.

Baseline Descriptives of Study Sample (N=199) overall and by experimental arm

	Overall Study Population (n=199)	Enhanced Intervention Arm (N=102)	Original Intervention Arm (N=97)	p-value
Age (years)	43.83 (10.11)	43.55 (10.63)	44.14 (9.57)	0.68
Generation				0.58
First	162 (82%)	83 (82%)	79 (82%)
Second	34 (17%)	18 (18%)	16 (17%)
Third	1 (.5%)	0	1 (1%)
Speak only Spanish or more Spanish than English at home (%)	157 (80%)	80 (79%)	77 (81%)	0.15
Country of Origin				0.30
Puerto Rican	1 (0.5%)	1 (1%)	0
Mexican, Mexican American, Chicana	177 (89%)	92 (90%)	85 (87%)
Cuban	1 (.5%)	1 (1%)	0
Guatemalan	1 (.5%)	0	1 (1%)
Colombian	2 (1%)	1 (1%)	1 (1%)
Other	19 (9.5%)	9 (9%)	10 (10%)
Education				0.15
<12 yrs	56 (29%)	24 (24%)	32 (33%)
HS Grad	30 (15%)	12 (12%)	18 (19%)
Vocational/Technical	29 (15%)	20 (20%)	9 (9.4%)
Some College	35 (18%)	19 (19%)	16 (17%)
College Graduate	34 (17%)	20 (20%)	14 (15%)
Post-Graduate Studies	12 (6.1%)	5 (5%)	7 (7%)
Employment				0.07
Unemployed	84 (43%)	37 (37%)	47 (50%)
Full Time	53 (27%)	34 (34%)	19 (20%)
Part Time	59 (30%)	30 (30%)	29 (30%)
Income (N=197)*				.002
<$20,000	82 (42%)	32 (32%)	50 (52%)
$20,000-29,999	36 (18%)	27 (27%)	9 (9%)
$30,000-39,999	28 (14%)	16 (16%)	12 (13%)
$40,000-49,999	16 (8%)	11 (11%)	5 (5%)
>=$50,000	26 (13%)	10 (10%)	16 (17%)
Don’t Know	9 (5%)	5 (5%)	4 (4%)
Marital Status				0.46
Never Married/Partnered	30 (15%)	16 (16%)	14 (15%)
Divorced or Separated	40 (20%)	24 (24%)	16 (17%)
Widowed	4(2%)	3(3%)	1(1%)
Married or Partnered	123 (63%)	59 (58%)	64 (67%)
Baseline min/week Objectively Measured MVPA	39.51 (71.20) median = 10	33.61 (71.29), median = 0	45.72 (70.93), median = 10.00	0.23
Baseline min/week of Self-reported MVPA	12.47 (22.54) Median = 0	14.33 (25.06), median = 0	10.51 (19.47), median = 0	0.23
Processes of Change
Cognitive (1-4.7)	2.61 (.87)	2.60 (.86)	2.62 (.89)	0.89
Behavioral (1-4.3)	1.99 (.67)	1.95 (.63)	2.02 (.70)	0.42
Self-Efficacy (1-5)	2.04 (.83)	2.04 (.91)	2.04 (.74)	0.99
Social Support
Friends (10-38)	14.58 (6.33)	14.58 (5.93)	14.58 (6.76)	0.99
Family (10-44)	16.72 (7.36)	16.67 (7.58)	16.77 (7.16)	0.93
Rewards and Punishments (3-8)	3.48 (1.04)	3.46 (1.04)	3.50 (1.05)	0.77
Enjoyment (PACES)* (31-126)	86.95 (20.18)	89.74 (20.14)	83.95 (19.89)	.04
Perceived Stress (3-40)	21.49 (8.55)	21.65 (9.00)	21.32 (8.09)	0.79
Depressive Symptoms (CES-D) (0-27)	7.76 (5.78)	7.97 (6.34)	7.53 (5.13)	0.59
BMI	30.43 (5.22)	30.50 (9.31)	30.79 (5.19)	0.33
Overweight/Obese (%)	168 (86%)	88 (86%)	80 (85%)	0.82
Waist Circumference	37.31 (4.97)	36.73(4.80)	37.95 (5.11)	0.09
Blood Pressure
Systolic	111.52 (13.92)	110.16 (14.03)	112.99 (13.73)	0.16
Diastolic	70.47 (9.07)	69.90 (9.49)	71.07 (8.61)	0.37
Resting Heart Rate	68.72 (9.03)	68.72 (9.03)	68.72 (9.07)	0.99
Percent Body Fat	39.53 (6.88)	39.41 (7.01)	39.66 (6.76)	0.80

Means (std deviations) are presented unless noted otherwise.

^*p<.05 for between-group differences. Ranges of scale scores presented in parentheses.

Discussion

The baseline data obtained from the Latina participants in this study show that they are very inactive, similar to patterns of inadequate physical activity in national samples, [5] and are at increased health risk due to the combination of overweight and low physical activity levels. The significant disparity in rates of lifestyle-related illnesses between Latinas and the general population [2–4] makes the need for intervention particularly urgent.

Comparisons with Seamos Saludables.

Our previous study, Seamos Saludables [29, 67], was conducted in New England with participants who were primarily Dominican, Colombian, and Puerto Rican, in contrast to the current study whose participants were almost exclusively Mexican/Mexican-American/Chicana. Participants in the present study were also slightly older (43.8 years, SD=10.11) than in our previous study (40.67 years, SD=9.98), however in other respects Latinas in both studies were very similar. A large majority of women from both Saludables (82%) and the present study (79%) spoke primarily or exclusively Spanish at home, and over 80% of women in both studies were first generation immigrants to the US. Rates of physical inactivity were likewise similar: women in both studies self-reported a median of 0 minutes per week of physical activity at baseline.

While Seamos Saludables provided a culturally and linguistically adapted print intervention that was effective for increasing physical activity, the Seamos Activas II study addresses additional challenges by targeting theory-derived constructs (social support, enjoyment and expectations) that are central to physical activity adoption, and by adapting delivery methods to include mobile technology (as text messaging) to increase the immediacy of the intervention and allow for greater accountability and interactivity. This study provides a rigorous test of the added value of addressing specific theoretical constructs and enhancing interactivity while building upon a proven-effective intervention.

The baseline data from the Seamos Activas II intervention shows room for improvement in PA-related theory-derived psychosocial measures including self-efficacy, processes of change, social support and exercise enjoyment. Such variables have been show to be mediators in previous studies [70, 76–81]. The original Seamos Saludables study found that family and friend support for PA increased more for the intervention group than the wellness control [70]. Family participation mediated the effect of the intervention on objectively measured MVPA at post-treatment whereas friend participation mediated the intervention effect on self-reported MVPA at the 12 month follow-up indicating that participants who had friends encourage and join them in PA were better able to sustain MVPA gains. The current Seamos Activas II intervention is building in more possibilities for social support through interactive texting.

The incorporation of text messaging into the present study responds to feedback obtained in our previous study in which Latinas requested greater interactivity and accountabilty within the intervention. These changes extend the reach of this intervention while maintaining the “always present/always available” quality of mobile interventions and respond to trends in mobile phone ownership, which are at near saturation levels for all demographic groups in the United States. [33, 34]. While text messaging strategies have yet (to our knowledge) been used for physical activity promotion in Latinas, promising results from recent studies in this area with other underserved/at risk populations (African Americans; [82, 83] low income rural women;[84] persons living in permanent supportive housing;[85] women with young children;[86] cancer survivors)[87, 88] as well as published formative research efforts to determine key features of text-messaging PA interventions for urban, low-income Latino patients with diabetes,[89] bode well for participant engagement and satisfaction in the current study.

Results from this study are also enhanced by the rigor of the overall study design and assessment methods. For example, providing participants with an in-vivo experience of moderate intensity physical activity (via treadmill walk) immediately prior to the PAR interview should be expected to enhance the accuracy of participant self-reports of MVPA. Self-reports are also bolstered by the use of actigraphy which provides an objective measure of MVPA.

Transferring the Seamos Activas II intervention to an even more cost-effective, user-friendly smart phone application may represent an important potential future direction. The majority of Latinos own smartphones, however there are disparities within the community as 87% of U.S. born Latinos own a smartphone, but only 67% of foreign-born Latinos do [90]. Thus, it may be premature to move our intervention into the smartphone space needed for app development, given that many of who need it most would be left behind. However, the digital divide continues to shrink, which is favorable for the m-health future directions for this line of physical activity-related health disparities research in Latinas.

In summary, Seamos Saludables II is a low cost, high reach, theory and technology-enhanced intervention that will be vetted across an unprecedented range of Latino subgroups. We anticipate that participants randomized to the Enhanced Intervention arm will show significantly greater increases in minutes of MVPA at six and twelve months than those in the Original Intervention arm. If the aims are achieved, our intervention will be poised to provide a more effective and highly disseminable home-based approach to achieve national PA guidelines and reduce related health disparities in this at-risk population.