The SMART trial is a single-center, 24-month clinical trial of adults seeking treatment for weight loss. The study uses a 3-group design to determine the effects of different modes of self-monitoring on short- and long-term weight loss and on adherence to self-monitoring. Participants are randomized to one of three conditions: (1) PPR, (2) PDA with dietary and physical activity software, and (3) PDA+FB with the same software plus a customized feedback program.
This study was approved by the University of Pittsburgh Institutional Review Board. All participants provided informed consent.
The eligibility criteria are listed in . Participants needed to be overweight or obese but otherwise in good health, and participants needed to be reasonably certain that they would be able to attend the treatment sessions and assessments throughout the two-year study. Eligibility was assessed several times throughout the screening process as described below.
Inclusion and exclusion criteria for the SMART trial.
2.2. Recruitment, screening and enrollment
Participants were recruited from the general population in the Pittsburgh area in three annual cohorts (2006, 2007, and 2008). Several recruitment strategies were used to increase the representativeness of the sample. Participants from the principal investigator’s previous clinical trial, PREFER, were asked to refer friends and family to the SMART trial. We also maintained a weight research registry of individuals who expressed an interest in receiving information about future studies. Participants in the registry were sent a letter or e-mail describing the study. Those who were sent a letter were given an extra study flyer to share with a friend. The university and adjacent medical center both have voice mail announcement systems that were also used to recruit participants. Flyers were placed throughout the university, medical center and community. Mailing lists were purchased for recruitment of the second and third cohorts. An advertisement was placed on City of Pittsburgh buses for recruitment of the third cohort.
Participants were evaluated through a multi-stage screening process. Refer to for the number of participants at each stage. The first stage of screening occurred over the telephone. Individuals calling in response to advertisements were asked a series of screening questions to determine basic eligibility. If an individual remained eligible he/she was sent a packet of three questionnaires to collect general demographic information, determine if the individual had a binge eating disorder and assess health history including medication use and alcohol consumption. If the individual scored above 36 on the Binge Eating Scale, which assesses disordered eating habits, the person was excluded from participation and given information on treatment sources [31
]. Individuals who remained eligible after the review of their screening questionnaires were invited to an orientation session. At this session, which lasted approximately 1 h, the participants were measured to verify their self reported weight and height, the study was explained to them in detail, they had the opportunity to ask questions, and they were given a 5-day food diary. The diary served two purposes, 1) to provide participants a time to try self-monitoring and decide if they wanted to commit to a two-year study of self-monitoring and 2) to screen out individuals who would not be able to self-monitor or who demonstrated that they would not complete the diaries as instructed. A baseline assessment was scheduled for each participant who returned a completed diary.
2.3. Baseline assessment
The baseline assessment included completing 2 unannounced 24-h dietary recalls (1 work day and 1 leisure day), a battery of questionnaires and a visit to the Clinical Translational Research Center (CTRC). The questionnaires were given or sent to the participants in advance of their assessment appointment and they were asked to bring the completed materials to the appointment. At the appointment, staff immediately reviewed all questionnaires for incompleteness and asked the participant for clarification if needed.
After completing a 12-h overnight fast, participants came to the CTRC for anthropometric (height, weight, body composition, and waist circumference) and blood pressure measurements, and had their blood drawn for lipid and glucose measurements. Participants with elevated glucose levels (>125 mg/DL) were not eligible to continue in the study. All other individuals who completed the assessment were eligible for randomization.
Once we confirmed that all participants for a given cohort were eligible, we proceeded with randomization. Individuals were stratified by gender and ethnicity (White vs. non-White) to ensure balance across the three treatment groups.
shows the data collection schedule. Participants were asked to complete assessments that were similar to the baseline assessment every 6 months. Due to budgetary constraints lipid and glucose assays were conducted only at the baseline, 12-month and 24-month assessments. Participants were compensated for their time at the 6-, 12-, 18-, and 24-month assessments.
Schedule of data collection.
2.5.1. Biological and physical measures
Venipuncture for the collection of serum for the lipid profile (total cholesterol, high density lipoprotein, low density lipoprotein, very low density lipoprotein and triglycerides) and glucose was performed by a trained phlebotomist with the subject sitting upright following a 12-h fast and a brief resting period. A wall-mounted stadiometer was used to measure height and a Tanita scale and body fat analyzer (Tanita Corporation of America, Inc., Illinois, USA) were used for weight and body compositions, which were assessed via bioelectrical impedance analysis. BMI was computed using the following formula: BMI=weight (kg)/height (meters2). Waist circumference was measured at least twice with a Gulick II measuring tape. When two measurements were obtained that were within 2 cm of each other, the average of the two measurements was used. Blood pressure was measured following the American Heart Association standard guidelines with the subject in a sitting position after at least a 5-min rest period, using an automatic blood pressure monitor.
2.5.2. Measures of treatment adherence
Adherence to the dietary protocol was assessed through two 24-h dietary recalls, which were conducted using the Nutrition Data System for Research Software (NDSR). At the orientation sessions, individuals were given a booklet of measurement figures to keep near their telephone so they could use this for estimating portion size when they received the phone call for the dietary recall. They were also asked to complete a form that identifies which phone number (mobile, home or work) to use and the best times for the interviewer to reach them to conduct the recall. The two calls were made unannounced, one on a work day and one on a leisure day. The interviews were conducted using the Five-Step Automated Multiple-Pass Method [32
] and the NDSR software to guide the food recall process. The NDSR is a comprehensive nutrient calculation software program that is maintained by the Nutrition Coordinating Center at the University of Minnesota and is updated annually to reflect marketplace changes and new analytic data. The NDSR database contains over 18,000 foods, 7000 brand names and an increasing number of ethnic foods. It also features optional dietary supplement data that we include with 24-h dietary recalls. These data provide total calories, fat grams and values for 156 nutrients with 9 food groups and 166 subgroups for our analysis. A member of the team who has a master’s degree in nutrition science has been trained at the University of Minnesota in the use of the NDSR program; she either conducts the recalls or supervises dietetic students whom she has trained.
The Modified Activity Questionnaire[34
] assessed daily and leisure activities and provided a measure of adherence to the physical activity protocol. Participants also recorded the number of steps that were recorded on their pedometer in their diary.
Additional measures of adherence included attendance at the treatment sessions and recording of food and exercise behaviors either in the PRs or PDAs. PDAs were uploaded at each treatment session and the diary content printed to replicate the paper record for the interventionists’ review and to assess completion of the diary. The software in the PDA (Dietmate Pro©[36
] and CalculFit©, PICS, Reston, VA) date- and time-stamped each entry in the PDA, permitting us to examine adherence to self-monitoring as well as temporal patterns of adherence to self-monitoring over the course of the 24-month study.
2.5.3. Measures of factors related to adherence
Questionnaires evaluated potential moderators and mediators of adherence such as barriers to healthy eating, self-efficacy, problem solving, stress and the effects of life events on weight. Additional questionnaires addressed the participants’ experiences with and acceptance of technology. These measures included Barriers to Healthy Eating, a 22-item questionnaire in which subjects used a scale of 1 (no problem) to 5 (very important problem) to rate various situations or conditions related to following the diet, e.g., feelings of deprivation and cost of the regimen [38
]. A higher score indicates more barriers. Experiences Associated with Following a Low-Fat diet (ELF) is a 25-item scale that assessed experiences thought to be related to diet maintenance. The scale had a total of 6 factors: wellness, distaste (for fat), cost, inconvenience, deprivation, and family and asked about past experiences; therefore, it was administered at 3 months rather than at baseline. The response scale was 1 (strongly disagree) to 5 (strongly agree) with a higher score meaning a more positive experience and attitude related to following the low-fat diet [39
]. The Medical Outcomes Study, Short Form-36 Questionnaire (SF-36), was being used to measure general health-related quality of life, how it was affected by weight reduction, and its impact on adherence to the lifestyle change. The scale provided two summary scores (Physical Component Score and Mental Component Score) with a range of 0 to 100 [40
]. The General Attitudes Toward Computers Scale (GATCS-C) consisted of 20 items that used a 5-point Likert scale to measure attitudes toward technology [41
]. We used this instrument because a person’s attitude toward technology could influence success in using the PDA; it could also change with use over time.
2.5.4. Measurement quality control
Data were collected at the CTRC by trained and certified staff using standardized procedures and questionnaires. Equipment was standardized and routinely calibrated. Questionnaires were checked for accuracy and completeness prior to the participant leaving the center. Each staff person who oversaw or completed an assessment procedure recorded his or her initials on a form that documented each assessment procedure. The form was reviewed and initialed by the participant at the end of the appointment.
2.6. Intervention sessions
Group meetings were held in the evening and lasted approximately 45 to 90 min. They were held weekly for the first 4 months, then every two weeks for the remainder of the first year. In the second year group meetings were held monthly for months 13–18 and then once in month 21, halfway between the 18- and 24-month assessments. Participants were weighed at the beginning of each group session in a private office. At the 39 group sessions they received nutritional and behavioral counseling and practical hands-on experiences to develop skills to implement a healthy lifestyle. A detailed listing of the central features and strategies of the intervention has been published previously [42
]. In response to comments received during an evaluation of a previous study, every fourth session was a supervised exercise session where participants were able to experience a variety of activities such as yoga, indoor and outdoor walking, and strength training with supervision and receive feedback on performance provided by the lead interventionist, a master’s level prepared exercise physiologist.
Participants received a calorie goal based on their baseline weight and gender (women: 1200 kcal for <200 lb or 1500 kcal for ≥200 lb; men: 1500 kcal for <200 lb or 1800 kcal for ≥200 lb) and were asked to limit their fat intake to no more than 25% of their calories, e.g., 33 or 42 g per day for females.
2.7. Treatment assignment
All participants received the same standard intervention; the only difference was in the mode of self-monitoring that participants were asked to use. Participants in the PR group were asked to self-monitor using standard paper diaries. Participants in the PDA and PDA+ FB group were provided with Palm Tungsten E2™
PDAs and self-monitoring software (Dietmate Pro©[36
] and CalculFit©, PICs, Reston, VA). Participants in the PDA+FB group had additional custom software loaded onto their PDAs that tracked their adherence to calorie, fat gram and exercise goals and provided a daily message related to goal attainment.
Participants in the PR group were provided with a reference book with nutrition information. The first cohort was given the T-factor book [43
] and the second and third cohorts were given the Calorie King book [44
]. These participants were trained in how to look up the nutritional information for their foods at the first group session, and they were given a pen and calculator set with the study logo.
Participants in the PDA and PDA+FB group were trained how to use a PDA and how to use the self-monitoring software during the first two group sessions. The training included the use of PowerPoint slides and the use of a PDA simulator projected on the screen so all participants could see the functions of the PDA software and work along with the instructor. Adequate mastery of the basic functions for self-monitoring was achieved in the first two sessions for most participants. One-on-one assistance was provided for participants who needed additional help.
Self-monitoring was the focus of this trial. Participants were asked to self-monitor their eating and exercise behaviors in a timely fashion throughout the 2-year study. The importance of timeliness in recording intake and subtotals of calories and fat consumed was stressed because it gives participants a chance to adjust their behavior during the day to avoid exceeding their calorie or fat gram goals. The PDA automatically calculated the subtotals after each food entry whereas the subtotals needed to be calculated for the PR. At each session, PR participants turned in their diaries and received new diaries to use until the next session. They also received the diaries turned in at the previous session that had been reviewed by and contained feedback from the interventionists. When the meeting frequency decreased to less than weekly, the diaries for the past two weeks were reviewed; however, if a person missed a session and then turned in four diaries, only the last two were reviewed and feedback was provided by the interventionists. The PDA and PDA+FB participants turned in their PDAs at the beginning of the session. During the session the self-monitoring data from the participants’ PDAs were uploaded to the PICS website and converted to a format that could be downloaded to the study’s Microsoft (MS) Access Adherence database. The interventionists then received printed reports that looked similar to the standard paper diaries for their review and comments. These annotated printouts were given to the participants at the next session with feedback. Multiple weeks and missed sessions were handled as described for the PR group.
2.9. DietMatePro and Calcufit software
The DietMatePro© (DM Pro) software contains a USDA-based database of nutrient information for over 6000 food items. Moreover, the database contains nutritional information for foods from many national restaurant chains and participants have the option of adding unlisted foods to the database. Besides being able to add individual food items, participants also can add entire meals to the database. For example, if a participant always had a cup of coffee, a bowl of cereal with skim milk, and a medium banana for breakfast, those items could be entered as “Breakfast 1” then each morning the participant could simply enter “Breakfast 1” into the PDA instead of entering each individual food item.
The Calcufit software allowed participants to enter the intensity of and amount of time in aerobic exercise, the amount of weight and number of repetitions of strength training exercises, and details about stretching, as well as steps counted and water consumed. As with DietMatePro, the date and time of each entry were recorded, and no forward entry was allowed with either software.
The DM Pro developer created a custom desktop application that facilitated the transfer or upload of data from the PDA to the web site (hotsync) and server of DMPro and Calcufit. During uploading, this application copied all food and exercise records from the PDAs to the server hosted by the software developer. If there was a technical failure of a PDA and the data were lost, previous records could be brought back to the PDA from the web server during the upload process. Moreover, back-up software (BackupMan) was loaded onto the PDAs and set to back-up data to secure digital memory cards every night, further reducing the loss of data that would have occurred during a device failure.
2.10. Customized feedback program
A customized feedback program was designed for this study. The algorithm for the dietary goal feedback was written in four groups corresponding to time of day with five sections in each group corresponding to goal attainment. displays the criteria for each group/section combination. The program randomly selected a time of day such that each participant received one dietary feedback message each day. This randomization was weighted to reduce the likelihood of the participants receiving a message at the same time on (the same part) on two consecutive days. The algorithm intentionally excluded some participants. We did not think that a feedback message was appropriate for participants who were overly restrictive in their eating. Presumably these participants were identified by the interventionists and were offered appropriate counseling. It was impossible to distinguish in an algorithm those participants who were engaging in restrictive eating from those who only partially recorded their intake, but we thought that the most prudent action was to allow the interventionists to explore these issues in person. We also counted participants as adherent if they slightly exceeded their goal (101% of calorie goal or 102% of fat gram goal). This rule took into account the imprecision of calorie and fat gram counting and avoided discouraging participants who simply appeared to exceed the goal due to rounding.
Feedback algorithm for calorie goals.
The exercise algorithm was much simpler because measuring adherence to exercise is much more straightforward than is adherence to the dietary goals. Since exercise goals were given as weekly minutes to achieve, the weekly goal was divided by 7 to determine an average daily goal. The exercise messages were delivered once every other day since participants were not expected to exercise every day but were encouraged to exercise on most days. Thus, the average daily goal was multiplied by 2 and the participants’ exercise minutes for the previous 2 days were compared to the 2-day goal. Participants could receive a message between 3 and 5 PM or between 7 and 8 PM. Each group of messages contained three sections: 1) the participant did not record, 2) the participant met or was likely to meet the goal (minutes were equal to or greater than 80% of the goal), 3) the participant did not meet or was not likely to meet the goal (minutes were equal to or less than 50% of the goal).
The algorithm remained the same throughout the study, but the library of messages was changed after a year to reduce the risk of the participants becoming desensitized by repeatedly seeing the same messages throughout the two-year study.
2.11. Data and safety monitoring plan
Because this was a low risk study, a data and safety monitoring committee was not required; however, we had a detailed plan and a Data and Safety Monitoring Officer (DSMO). The primary safety concerns were gastrointestinal and musculoskeletal. A Symptom and Injury Survey was developed for the study and was administered at each assessment appointment. Biological measures were routinely collected and reviewed for safety monitoring, e.g., lipid profile, glucose and blood pressure. Results were reviewed by the PI and shared with the DSMO, a physician with extensive experience in weight management. If any of these parameters were elevated, the participant was advised to inform his or her physician. All participants were provided a written report of the biological and physical measure results from each assessment.
2.12. Data management
The study used a password protected centralized Oracle 9i database for storage of the outcome data, which were identified by subject identification number only. The outcome data were collected on forms designed to permit scanning using Teleform 6.1 software. Data were verified for accuracy after they were scanned into the database. Visual Basic 6.0 program was used to clean all scanned data by employing logic for data correction, range checks, skip and fill, coding missing values, and coding non-applicable questions. The progress of all participants was tracked throughout the study in a MS Access tracking program that permitted the staff to query the system and develop reports at any time. Process data (completion of weekly diary, achievement of dietary and physical activity goals, weights and attendance) were entered weekly in a MS Access database so retention and progress could be reviewed monthly. Status reports of screening, recruitment, and assessments were provided frequently. Only trained project staff were permitted to scan or clean the data and first had to provide a logon name and password.
2.13. Analysis plan
The primary outcome measure is weight change from baseline to 6 months and baseline to 24 months. The design of the study required data collection at baseline, 6, 12, 18, and 24 months. Therefore, missing data may present a problem in the data analyses. Analyses will follow the intention-to-treat model (ITT), that is, all randomized participants will be included in the analyses according to their original assignment. Individuals who withdraw or do not complete the final assessment will be included in the analysis by using the baseline weight as imputed values; in other words, no weight change will be assumed. The ITT approach will tend to bias the analyses in the direction of the null hypothesis of no difference among groups. We will examine the cohort effect in the analyses.
2.14. Sample size justification
The sample size for this study was determined considering 1) the effect sizes observed in the PI’s preliminary work for changes in the primary outcome of weight from baseline to follow-up (6, 12, and 18 months) across the estimated levels of self-monitoring (>25% to
50%, >50% to
75%, and > 75% to
100%); these effect sizes are expected to parallel the differences in weight loss across the treatment groups (PR, PDA, PDA+FB) with the level of significance adjusted to .01 to accommodate multiple testing; and 2) the expected rate of attrition during the 24-month study period. Taking these two factors into consideration we determined that a final sample size of 156 participants would be required to adequately test our hypotheses. We anticipated that we would need to enroll 198 individuals to retain 156 participants at 24 months.