Effect of Telehealth Extended Care for Maintenance of Weight Loss in Rural US Communities

Key Points Question Does extended care for obesity management delivered remotely via individual or group telephone counseling in rural communities improve the maintenance of weight loss compared with an education program about weight gain prevention? Findings In this randomized clinical trial of 445 participants who completed a lifestyle weight-loss intervention, extended care delivered remotely via individual telephone counseling decreased weight regain and increased the proportion of participants who achieved clinically meaningful weight losses compared with education alone. Meaning These findings suggest that extended care delivered remotely via individual telephone counseling may reduce weight regain and increase the proportion of participants who achieve long-term weight reductions of at least 10%.

with a bachelor's or master's degree in nutrition, exercise science, or psychology. All 132 interventionists will be provided with training in lifestyle treatment and nutrition education that 133 includes two 8-hour workshops held semi-annually, plus case management reviews with a 134 clinical health psychologist (MGP) or registered dietitian (MNS), each with extensive experience 135 overseeing behavioral interventions for obesity. The case management reviews are conducted 136 weekly during Months 1 through 4, biweekly during Months 5 through 10, and monthly during 137 Months 11 through 16. Interventions sessions are audio-taped and random reviews are carried 138 out to ensure treatment fidelity. Corrective feedback is provided during training, case 139 management reviews, or individually with the interventionists. 140

Phase I Lifestyle Intervention 141
Phase I will consist of 16 weekly face-to-face group sessions conducted at local CES 142 offices. Each group will consist of 4-16 participants, and each session will last approximately 90 143 minutes. Intervention content was derived from the Diabetes Prevention Program (DPP) 144 [42,43]. Modifications to the DPP approach include group rather than individual counseling [44] 145 and home-based rather than center-based physical activity [45]. The DPP intervention was 146 further modified to address weight-management challenges experienced by individuals living in 147 rural areas (i.e., lack of places to exercise, traditions of high-fat Southern cooking, absence of 148 social support for weight loss, etc.). Similar to the DPP, the Rural LEAP intervention is guided 149 by social-cognitive and self-regulation theories [46][47][48], and it incorporates key behavioral 150 modification strategies including goal setting, self-monitoring of dietary intake and physical 151 activity, stimulus control, cognitive restructuring, and problem solving [33]. Utilization of 152 problem-solving skills, in particular, is emphasized as a way to address the specific challenges 153 experienced by adults living in rural communities [49,50]. Topics and objectives for the 16 154 Phase I and 18 Phase II weight-loss treatment modules are presented in Table 2. 155 The goals of the initial lifestyle intervention are to decrease caloric intake in a 156 nutritionally sound manner [51-53] so as to produce a weight loss of approximately 0.4-0.9 kg 157 per week and to increase home-based walking to 210 minutes per week [54]. Initial caloric 158 intake goals are determined by the participant's baseline weight (e.g., 1200 kcal/day for 159 participants weighing ≤ 113.6 kg; 1500 kcal/day for those weighing > 113.6 kg). Participants will 160 be instructed to keep daily logs of their food and drink intake, including the types and amounts 161 of foods consumed along with corresponding caloric values, using either written logs or online 162 programs/applications (e.g., MyFitnessPal ® or Lose It! ® ). Participants will also be instructed to 163 keep daily records of their steps using a pedometer or activity monitor. To support these self-164 monitoring efforts, participants will be provided, at no cost, with a Calorie King reference book 165 [55,56], blank food logs, measuring cups, a food scale, and a clip-on pedometer. As a 166 substitute for the use of a pedometer, participants will be permitted to self-monitor physical 167 activity via self-purchased electronic tracking devices (e.g., Fitbit ® , Garmin ® , or Apple Watch ® ). 168 Participants will be encouraged to work toward meeting their daily caloric intake goals by 169 selecting lower-calorie food options, decreasing consumption of "fast food," decreasing or 170 eliminating sugar-sweetened beverages, increasing consumption of fruits, vegetables, and 171 whole grains, and using lower-fat methods of preparing foods. Weight-loss treatment modules 172 included instruction on the USDA's MyPlate to emphasize choosing whole grains, fruits, 173 vegetables, lean meat, and low-fat dairy products [51]. 174 Walking will be the primary form of prescribed physical activity. During week 2, each 175 participant's steps will be monitored over 7 days, and the participants will then be encouraged to 176 gradually increase their planned daily walking time by 5-10 minutes (i.e., 500-1000 steps) each 177 week as tolerated until they reach an average of 30 minutes/day (3000 steps) above their 178 baseline level. Participants who achieve this objective will be encouraged to further increase 179 their steps toward an ultimate goal of 60 min/day of walking for exercise (i.e., 6000 steps above 180 baseline). 181

Phase II Randomization and Extended-Care Lifestyle Intervention 182
To maximize participation in the randomized and follow-up phases (II and III) of the trial, 183 only those individuals who attend fewer than 50% of the Phase I sessions will be excluded from 184 advancing to Phase II randomized assignments. All other participants, regardless of initial 185 weight loss achieved, will be randomly assigned to one of the three Phase II conditions (GRP,186 IND,or CTRL). Randomization was completed by the study statistician (MJD) using the random 187 number generator in R statistical computing software [57]. County size, county, Phase I group, 188 and session time will be balanced during randomization. Interventionists will notify participants 189 of their Phase II group assignment during the last Phase I session. The primary objectives of 190 Phase II (Months 5-16) are to sustain the lifestyle changes accomplished in Phase I so as to 191 maintain lost weight and to make further adjustments in energy balance to produce additional 192 weight loss during "campaigns" conducted in Months 8-9 and Months 13-14. 193 The mode of extended-care contacts will differ by condition during Phase II, but the 194 schedule of contacts is consistent across the conditions. Intervention contact across all three 195 conditions will occur bi-weekly during Month 5 through 10 and monthly during Months 11 196 through 16. At each of the scheduled contact points, participants in all three conditions receive 197 a weight-loss treatment module delivered via e-mail or U.S. mail with information and 198 recommended behavioral activities targeted at maintaining lost weight. All participants will also 199 be provided with a supply of blank self-monitoring logs and pre-paid mailers for returning 200 completed dietary logs to the interventionists. 201

GRP condition 202
Participants randomized to the GRP condition will call into a teleconference line for 60-203 minute group sessions with the interventionist and other members from their Phase I group. 204 The calls will include three primary components: (1) a group "check-in" during which participants 205 report progress toward goals and problems experienced since the previous session coupled 206 with group problem solving of one or two complex issues to address barriers to progress [49]; 207 (2) group discussion of the treatment module with the interventionist explaining new objectives 208 and suggesting behavioral strategies to achieve those objectives; and (3) structured individual 209 goal setting with respect to each participant's plans for diet and/or physical activity changes to 210 be accomplished prior to the next session. To maximize the time efficiency of the group 211 session, all participants will be asked to identify their calorie and step goals prior to the call. 212 The group problem-solving activities facilitated by the interventionist during the 213 extended-care sessions will incorporate the use of an evidence-based, five-step model [49,50] 214 including: (1) orientation (i.e., developing an appropriate coping perspective --"Problems are a 215 normal part of managing your weight, but they can be dealt with effectively."); (2) definition (i.e., 216 specifying the problem and goal behaviors -"What is the particular problem facing you right 217 now? What is your goal in this situation?"); (3) generation of alternatives (i.e., brainstorming 218 potential solutions -"The greater the range of possible solutions you consider, the greater your 219 chances of developing an effective solution."); (4) decision making (i.e., anticipating the 220 probable outcomes of different options -"What are the likely shortand long-term 221 consequences of each of your options?"; and (5) implementation and evaluation (i.e., trying out 222 a plan and evaluating its effectiveness -"What solution plan are you going to try and how will 223 you know if it works?"). 224

IND condition 225
Participants randomized to the IND condition call into a teleconference line at a 226 predetermined appointment time for a one-on-one, 10-to 20-minute session with the 227 interventionist. Participants will be contacted to schedule make-up calls in the event of time 228 conflicts or missed calls. Aside from the difference in call length and the absence of other group 229 members, IND calls will include the same structure, components, and problem-solving focus of 230 the GRP calls. 231

CTRL condition 232
Participants in the CTRL group will receive a series of 18 written educational weight-loss 233 treatment modules delivered via e-mail and/or U.S. mail delivered on the same schedule as the 234 GRP and IND contacts during Phase II. The modules will include the identical content used in 235 the GRP and IND conditions. The CTRL condition will not include any scheduled phone or in-236 person interactions with the interventionists or other participants. 237

Assessment visits 239
Study personnel masked to participant randomization will conduct in-person assessment 240 visits at baseline and at Months 4, 10, 16, and 22. Assessment visits will be conducted at the 241 same local CES site for all time points. Self-reported medical history and a list of medications 242 will be updated at each assessment visit. 243

Anthropometrics 244
At baseline, height will be measured using a stadiometer (ShorrBoard ® ), with 245 participant's shoes removed. At baseline and subsequent assessment visits, weight will be 246 measured with a calibrated digital scale (Tanita BWB-800S), with participants in light indoor 247 clothing, pockets emptied, and shoes removed. 248

Fasting blood samples 249
Fasting blood samples will be obtained by the study nurse at baseline and at Months 4 250 and 22 and analyzed for metabolic and lipid profiles by Quest Diagnostics ® . 251

Resting blood pressure and heart rate 252
Blood pressure and heart rate will be measured at baseline and Months 4 and 22. 253 Measurements will be taken while the participant is seated with feet resting on the floor for five 254 minutes in a quiet room free of distractions. Resting systolic and diastolic blood pressure will be 255 taken three times, two minutes apart, using a Dinamap ® Automated Vital Signs Monitor with 256 appropriate cuff size; the second and third readings will be entered and averaged in the study 257 database. 258

400 Meter Walk Test 259
The 400 Meter Walk Test [58], a standardized measure of physical fitness commonly 260 used to assess individuals with chronic health conditions, will be completed at baseline, and at 261 Months 4 and 22 to assess fitness. 262

Participant questionnaires 263
The following questionnaires will be completed by the participants at baseline and at 264 Months 4, 10, 16, and 22. 265

Paffenbarger Physical Activity Questionnaire 266
Self-reported physical activity will be collected via the Paffenbarger Physical Activity 267 Questionnaire [59], which includes questions on distance walked, flights of stairs climbed, and 268 time spent in other sports, recreational, or fitness activities during a typical day or week. 269

Medical Outcomes Study 36-Item Short Form Health Survey (SF-36) 270
The SF-36 [60] is a quality-of-life measure that assesses the following domains:

Social Provisions Scale 278
The 24 Florida Clinical and Translational Science Institute. Self-report data will be entered directly by 301 participants into REDCap's online survey module. All other data will be entered by study 302 personnel according to the protocol and in compliance with the U.S. Health Insurance Portability 303 and Accountability (HIPAA) Privacy Rule. Data will be exported to a SAS dataset for statistical 304 analyses [69]. 305

Missing data 306
Data analyses for this effectiveness trial will be performed using an intent-to-treat 307 approach that includes all randomized participants. It will be assumed that individuals who 308 discontinue participation in the study prior to Month 22 regained weight, on average, at a rate of 309 0.3 kg per month after leaving the study, up to their baseline weight. This conservative 310 approach to imputing missing weight values has been employed by numerous prior studies 311 [34,35,70] and is consistent with published reviews of weight regain following lifestyle treatment 312 for obesity [18,71,72]. We will also consider missing at random. Sensitivity to the assumptions 313 about the missing data [75] will be examined in our final analysis. 314

Sample size justification 315
The primary outcome will be change in body weight from Month 4 to Month 22, 316 comparing the GRP and IND phone-based conditions to the education CTRL condition. It is 317 assumed that the county size, county, and session time effects will be negligible due to the 318 balance in the randomization assignments. The analysis will, however, take into account a 319 random effect that allows for correlation among individuals in the same GRP sessions. Based 320 upon prior studies [35,76], a 20% attrition rate is expected by Month 22. The recruitment target 321 includes a total of 540 individuals entering Phase I (180 per condition). Power calculations were 322 computed using the expected attrition rate and a Bonferroni adjustment to account for 323 comparisons of the three treatment conditions. The study is expected to have > 80% power 324 (two-sided tests, type I error rate of 0.05) to detect a 2.5 kg difference in regain from Month 4 to 325 Month 22 between GRP vs. CTRL and IND vs. CTRL. 326

Primary analyses 327
Participants' weight change will be calculated as the difference between weight in 328 kilograms at Month 22 and weight at Month 4. The primary hypothesis is that both the GRP and 329 IND telephone conditions will produce greater weight reductions at Month 22 than the education 330 CTRL condition. To test this hypothesis, we will employ the NiNBayes R package 331 (github.com/theodds/NiNBayes), which models the weights longitudinally using a Dirichlet 332 process mixture of models.
[77] Treatment condition is randomly assigned. Two primary 333 contrasts of interest will be tested for significance: GRP vs. CTRL and IND vs. CTRL. 334 Bonferroni corrections will be used to control for type I error. In addition, percent change in body 335 weight during months 4 to 22 will be analyzed by computing percent changes in body weight 336 from the models used for the primary outcome, weight change. 337

Secondary analyses 338
For the secondary aim, the outcomes of interest are the proportions of participants within 339 a treatment condition who lost ≥ 5% and > 10% of body weight at Month 22. It is hypothesized 340 that both the GRP and IND conditions will result in greater proportions of participants with ≥ 5% 341 and >10% losses at Month 22 than the CTRL condition. Two contrasts will be tested for 342 significance: GRP vs. CTRL and IND vs. CTRL. Based on the Bayesian approach for the 343 primary outcome and missingness, for each contrast, we will test the null hypothesis that the 344 relevant proportions are equal versus the alternative that they are not equal. Process measures will include intervention-related activities (e.g., attendance, self-monitoring) 351 and participants' evaluation of program components and interventionist effectiveness. These 352 endpoints are continuous variables, and analyses will be the same as that described for the 353 primary endpoint. No adjustments will be made for multiple comparisons due to the exploratory 354 nature of these analyses. 355 2.9.6.2. Mediator analysis. Adherence to caloric intake goals will be evaluated as a 356 mediator of the relationship between the intervention conditions and weight change at Month 22. 357 Causal mediation analyses will be conducted using the mediate function from the mediation R 358 will be tracked from the service provider perspective [35]. Cost effectiveness will be measured 362 primarily as "dollars per kg loss per treatment condition" [80]. Cost effectiveness will also be 363 measured according to the recommended clinical cut-points of maintaining ≥ 5% loss of initial 364 weight [15]. Similar effectiveness for the GRP and IND conditions are expected, with lower 365 effectiveness in the CTRL condition. Cost effectiveness will also be assessed based on Quality-366 Adjusted Life Year (QALY) gains related to weight change and will be calculated using data 367 from the SF-36 questionnaires completed at baseline, Month 4, Month 16, and Month 22 to 368 derive a preference score using the Short Form-6 Dimension (SF6D) method; 10 items from the 369 SF-36 are used to calculate a QALY score [81]. 370 Incremental cost effectiveness ratios (ICERs) will identify the differential costs and 371 outcomes of switching from individual to group-based telephone counseling. A priori, it is 372 expected that the GRP condition will be less expensive than the IND condition but with similar 373 effectiveness. All costs will be adjusted to reflect constant 2017 dollars. Sensitivity analyses 374 will be conducted to determine whether costs and cost-effectiveness estimates are sensitive to 375 study-specific wage rates and other program costs. Extrapolation using sample median wage 376 rates or national wage data will be calculated and the analysis will be rerun to determine any 377 change in results. Additional sensitivity analyses will rely on either probabilistic models or 378 Eighth edition. Washington, D.C. 530 Table 1 Weight-loss treatment modules.