Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Dec 1.
Published in final edited form as: Health Psychol. 2020 Dec;39(12):1037–1047. doi: 10.1037/hea0000996

Patient-Provided e-Support in Reduced Intensity Obesity Treatment: The INSPIRE Randomized Controlled Trial

Tricia M Leahey 1, Tania B Huedo-Medina 1, Andrea Grenga 2, Linda Gay 2, Denise Fernandes 2, Zeely Denmat 1, Caroline Doyle 2, Remei Areny-Joval 1, Rena R Wing 2
PMCID: PMC8428547  NIHMSID: NIHMS1727594  PMID: 33252929

Abstract

Objective:

There are two types of patient supporters, peers (two individuals initiating health behavior change who support one another) and mentors (previously successful patient who supports incoming patients). Social comparison theory suggests that peers and mentors may elicit social comparison processes (patients may compare their progress to that of their peer/mentor), and these social comparisons could impact treatment outcomes. This fully powered randomized controlled trial is the first to examine the differential impact of peers and mentors on obesity treatment outcomes when added to reduced intensity treatment.

Methods:

Participants (N=278) were randomly assigned to reduced intensity behavioral weight loss treatment alone (rBWL), rBWL plus peer e-support (rBWL+Peer), or rBWL plus mentor e-support (rBWL+Mentor). rBWL involved periodic group sessions that decreased over time; when group sessions decreased, intensity of peer/mentor e-support increased. Weight and social comparison processes were assessed throughout the 12-month intervention.

Results:

There was a significant treatment effect; when group sessions became less frequent and peer/mentor e-support became more frequent, rBWL+Peer had significantly greater weight loss than rBWL alone, and rBWL+Mentor was not significantly different from the other two. Social comparison processes differed by treatment arm; rBWL+Peer participants tended to report more lateral social comparisons (my weight loss progress is “similar” to my peer’s) whereas rBWL+Mentor participants reported more upward comparisons (my weight loss progress is “worse than” my mentor’s). Upward comparisons were associated with poorer weight loss outcomes.

Conclusions:

Peer e-support may be an effective, low-cost, sustainable method for improving longer-term weight loss outcomes in reduced intensity obesity treatment.

Keywords: Peer support, obesity treatment, social comparison, lay support, weight loss

INTRODUCTION

The United States (U.S.) is currently experiencing an obesity epidemic. Over 39% of American adults have obesity (Hales, Carroll, Fryar, & Ogden, 2017). Excess adiposity is associated with serious health consequences, including increased risk for cardiovascular disease, type two diabetes, and certain types of cancer (Chu et al., 2018). As a result, it is anticipated that obesity will cause the first decline in life expectancy, with obesity associated medical conditions expected to cause up to a one year decline in life expectancy in the U.S. by the year 2050 (Stewart, Cutler, & Rosen, 2009; Olshansky et al., 2005). Moreover, obesity related-medical expenses place a tremendous burden on our healthcare system, costing the U.S. an estimated $147 billion dollars each year (Finkelstein et al., 2012).

To help address the epidemic of obesity, the Centers for Medicare and Medicaid Services (CMS) provide coverage for the behavioral treatment of obesity (CMS, 2011). Whereas evidence-based behavioral treatment (e.g., the Diabetes Prevention Program) commonly involves weekly sessions for the first four to six months, followed by bimonthly visits through month twelve, the CMS contact schedule involves a reduced intensity schedule: one month of weekly sessions, followed by bimonthly sessions for months two through six, and then monthly treatment from months seven through twelve. This is a 44% reduction in treatment sessions compared to traditional evidence-based approaches. While this reduced intensity approach is expected to yield clinically meaningful weight losses (Jensen, Ryan, Donato, et al., 2013), given findings that ongoing support from treatment providers improves obesity treatment outcomes (Perri et al., 1987), the CMS’s reduced intensity approach has raised concern that the contact schedule may be of insufficient dose for optimal weight loss outcomes (Carvajal, Wadden, Tsai, Peck, & Moran, 2013). Moreover, with increased emphasis on providing effective obesity treatment in primary care and community-based settings (Tsai, Remmert, Butryn, & Wadden, 2017; Fitzpatrick et al., 2016), efficient, low-cost options for improving obesity treatment outcomes are needed (Carvajal et al., 2013).

Adding patient-provided support (trained patients who provide emotional and informational support to one another) to otherwise reduced intensity interventions may cost-effectively improve outcomes. Given that fellow patients are facing the same health challenges, there is also evidence to suggest that patients provide one another with a unique sense of empathy, not duplicated by professionals, family, or friends (Dennis, 2003). There are two types of patient supporters: peers and mentors. Peer patient supporters are two patients initiating health behavior change who provide support to one another (i.e. bidirectional, or reciprocal, support) (Leahey & Wing, 2013). For example, two patients newly diagnosed with diabetes may support each other in lifestyle change. Given that these two individuals are newly focused on lifestyle change, they likely experience both successes and failures as they initiate and navigate health behavior change with one another. In contrast, mentors are patients who previously and successfully faced the health situation and provide support to incoming patients (e.g., individuals with well-controlled diabetes support individuals newly diagnosed with diabetes) (Leahey & Wing, 2013). Given their success, mentors have more mastery over health behavior change.

Peers and mentors have been shown to have differential effects on health outcomes. Peer patients have been shown to improve diabetes management behaviors, and, when added to active treatment, reciprocal peer support (two patients paired to support one another) has been shown to significantly improve six-month HbA1c outcomes compared to active treatment alone (Heisler, Vijan, Makki, & Piette, 2010). Interestingly, however, the impact of mentor patients (successful individuals who provide support to new patients) on health outcomes is mixed. While there is some evidence to suggest that mentors improve self-reported levels of physical activity (Castro, Pruitt, Buman, & King, 2011), other studies with objective outcomes have shown that adding mentors to treatment has no effect on health outcomes (e.g., HbA1c) (Keyserling et al., 2002). Given these findings suggesting that peer patient support may be more effective than mentor patient support, a preliminary pilot investigation examined whether support from peer vs. mentor patients differentially impacted weight loss outcomes during obesity treatment; results from this pilot study suggest that peer support produces better short-term weight losses compared to mentor support (Leahey & Wing, 2013). Taken together, there is some evidence to suggest that peer patient support may be more effective than mentor patient support for health behavior change; however, fully powered trials are needed that directly compare these two types of patient support. Furthermore, if such differences do indeed exist, it will be important to elucidate the underlying mechanisms of action.

Festinger’s Social Comparison Theory (Festinger, 1954) may provide insight into why peer support may be more effective than mentor support. People naturally and frequently compare their progress to relevant others (Festinger, 1954; Leahey, Crowther, & Mickelson, 2007). There are three types of social comparisons: downward (comparing to a “worse-off” other), lateral (comparing to a “similar” other), and upward (comparing to a “better-off” other) (Festinger, 1954). Downward and lateral comparisons have been associated with higher levels of motivation, encouragement, and greater health behavior change relative to upward social comparisons (Aspinwall & Taylor, 1993; Collins, 1996; Leahey & Gokee Larose, 2010; Lockwood, 1997; Tesser, Millar, & Moore, 1988). Moreover, upward social comparisons have been shown to induce frustration and hinder performance (Aspinwall & Taylor, 1993; Buunk & Ybema, 1997; Collins, 1996; Lockwood, 1997; Tesser et al., 1988). Given evidence that patients engage in social comparisons with one another (Parent & Fortin, 2000), and that peers naturally serve as “similar” comparison targets whereas mentors naturally serve as “better-off” social comparison targets, social comparison processes may be particularly relevant to patient-provided support. However, these potential mechanistic processes have never been systematically examined within the context of patient support. Further, other than in our pilot study, the two types of patient providers (peers and mentors) have never been directly compared. Comparing these two providers on health outcomes will both expand findings in the area of social comparison theory and inform clinical practice. That is, if peers are indeed more effective than mentors, given that peers are much easier to recruit than extremely successful patients, peers may be the more pragmatic approach to improve outcomes in clinical care.

The primary aim of the INtervention to Support Partners In Reaching Eating and exercise goals (INSPIRE) trial was to examine the efficacy of adding patient-provided support (peer and mentor) to an otherwise reduced intensity behavioral weight loss intervention. Participants were randomized to either: (a) reduced intensity behavioral weight loss treatment alone (rBWL), (b) reduced intensity behavioral weight loss plus peer e-support (rBWL+Peer), or (c) reduced intensity behavioral weight loss plus mentor e-support (rBWL+Mentor). The primary hypothesis was that rBWL+Peer would yield significantly better weight loss outcomes compared to rBWL+Mentor and rBWL alone over time. Secondary analyses examined whether, within rBWL+Peer and rBWL+Mentor, social comparison processes differed and mediated treatment outcomes; given that Peers are individuals actively trying to lose weight whereas mentors are successful weight losers, it was hypothesized that participants randomized to rBWL+Peer would report lower social comparison scores with their peer partner (consistent with more lateral comparisons) and that participants randomized to rBWL+Mentor would reporter higher social comparison scores with their mentor (consistent with upward comparisons) and that social comparison processes would mediate treatment effects. Further, given preliminary findings that being a weight loss mentor may improve mentors’ weight trajectories (Leahey & Wing, 2013), the effects of being involved in the intervention on mentors’ weight losses were explored. This is the first fully-powered trial to directly compare the efficacy of peer vs. mentor patient support. Results from this trial could provide important information regarding how best to improve treatment outcomes in reduced intensity programs with the use of patient-provided support and help to elucidate theory-based mechanisms by which peers and mentors impact health outcomes.

METHODS

PARTICIPANTS.

Participants were recruited through mass mailings, newspaper advertisements, and listservs. Evidence from the patient support literature suggests that similarities among patient supporters may be important for the development of cohesive, supportive patient relationships (Dennis, 2003; Hibbard et al., 2002); thus, the age and Body Mass Index (BMI) range of individuals enrolled was limited. Given that the majority of individuals who enroll in behavioral treatment for obesity are middle-aged, individuals 40–60 years of age and with a body mass index (BMI) of 30 to 40 kg/m2 were recruited. A BMI upper limit of 40kg/m2 was chosen as individuals with a BMI greater than 40 may require additional medical supervision. Exclusion criteria were: unwilling to engage in patient-provided support; scheduling conflicts; involvement in another weight loss program; taking weight loss medication; history of bariatric surgery; ≥ 5% weight loss during the 6-months prior to screening; pregnant, lactating, less than 6-months post-partum, or plans to become pregnant during the study period; reported heart condition, unable to walk two blocks without stopping, chest pain during periods of rest or activity, or loss of consciousness reported on the Physical Activity Readiness Questionnaire (Thomas, 1992) (physicians written permission was obtained for individuals reporting joint problems, medication use or other medical conditions that may limit ability to exercise); known medical condition that would jeopardize participant safety if involved in a weight management program with diet and exercise guidelines (e.g., cancer); and conditions that would render the participant unlikely to follow the study protocol (e.g., relocation, dementia, substance abuse). See Figure 1 for participant flow. Participants were recruited, received treatment, and were assessed from January 2013 through February 2017. The Miriam Hospital’s Institutional Review Board approved this study.

Figure 1.

Figure 1.

Open in a new tab

Participant flow.

RANDOMIZATION.

A total of 278 participants provided consent and completed final screening procedures, which included a behavioral run-in. As in other weight loss trials (e.g., DPP, Look AHEAD), the behavioral run-in for this trial involved self-monitoring of dietary intake for approximately one week and one-on-one discussion of study requirements with staff to ensure participant understanding and willingness to complete study procedures. Participants were randomly assigned by the trial biostatistician to one of three intervention arms: rBWL, rBWL+Peer support (participants in the same weight loss program were randomly paired to participate in reciprocal peer support), or rBWL+Mentor support (participants in this arm received support from a successful weight loser – someone who lost 10% of their initial body weight and kept it off for at least 1 year). Given that men tend to lose more weight than women, randomization was stratified by gender. In previous studies, an intraclass correlation (ICC) in weight losses among peer supporters and among participants assigned to the same mentor was observed (Leahey & Wing, 2013). Thus, to account for the ICC and associated variance inflation (Murray, 1998), a 5:6:6 treatment arm randomization scheme was used (rBWL : rBWL+Peer : rBWL+Mentor).

Randomization also occurred in the peer and mentor arms. Participants in rBWL+Peer were randomly paired within gender to one other individual in their weight loss group and trained (see below) to support one another. Similarly, participants in rBWL+Mentor were randomly paired within gender to a mentor (see below for training information); two or three incoming participants were assigned to one mentor. A peer supporter ratio of 1:1 and a mentor:mentee ratio of up to 1:3 was used for several reasons: peer supporters, each actively working towards health behavior change, may find it too burdensome and overwhelming to support more than one peer partner at a time, which could adversely affect treatment outcomes; the 1:1 peer supporter ratio has previously shown to yield optimal retention and treatment outcomes (Heisler & Piette, 2005; Heisler et al., 2010); and, in studies that have tested mentor supporters, those with lower mentor:mentee ratios (1:1 to 1:3) have superior mentor retention, superior participant satisfaction and retention, better outcomes, and fewer complaints regarding mentor burden compared to those with higher ratios (1:8 to 1:10) (Castro et al., 2011; Dorgo, Robinson, & Bader, 2009; Hibbard et al., 2002; Riegel & Carlson, 2004).

INTERVENTIONS.

Reduced intensity behavioral weight loss program (rBWL).

Instead of traditional treatment, which involves weekly treatment for 4–6 months followed by bimonthly treatment through month 12, the reduced intensity behavioral weight loss program herein had a tapered contact schedule: Phase 1 was 6 weeks of weekly group sessions, followed by Phase 2, which was 6 weeks of bimonthly group sessions, and then Phase 3, which was 9 months of monthly group sessions, for a total of 18 sessions over one year (vs. the traditional 36 sessions [e.g., DPP Research Group, 2002]). rBWL was delivered by a master’s level interventionist. Participants were given a weight loss goal of 1–2 pounds per week and an ultimate goal of losing 10% of initial body weight. To do this, they were provided calorie and fat gram goals depending on baseline weight (≤250lbs: 1200–1500 calories, >250 lbs: 1500–1800 calories, with 30% of calories from fat) and a physical activity goal of gradually increasing to 250 minutes of moderate to vigorous physical activity per week. Participants were privately weighed prior to each group session. Group session content was consistent with that of the Diabetes Prevention Program (DPP) and covered behavioral strategies to reduce calorie intake and increase physical activity, stimulus control, goal setting, problem solving, and relapse prevention (DPP Research Group, 2002).

Reduced intensity behavioral weight loss program plus peer support (rBWL+Peer).

The rBWL+Peer program was identical to the intervention described above with one exception, the addition of e-support from a peer (i.e. another individual in their weight loss group). That is, at program start, participants in this arm were randomly paired within gender to another participant in their weight loss group and engaged in reciprocal peer support with that individual throughout treatment, defined as providing support to and receiving support from their peer partner. In early group sessions, peer partners engaged in relationship building activities in which they got to know one another, identified things they had in common, and shared how they can help one another reach their weight loss goals. At the end of initial weekly sessions, peer participants were trained in how to effectively support their peer partner. Total training was approximately one hour in duration. Training focused on providing balanced feedback with use of the “sandwich technique” (supportive comment, followed by a constructive suggestion for meeting weight loss goals, and ending with a supportive comment) and evidence-based suggestions for weight loss (e.g., meal plans, pre-packaged meals). During training, peer partners were given worksheets with various peer progress scenarios (e.g., weight loss, weight gain, no weight change) and practiced providing support and suggestions; interventionists offered corrections/suggestions when needed.

During each week that there were no group sessions, peer partners exchanged their progress (weight change, average weekly calories, total activity minutes) with one another via email or text (participant preference; dyads had to agree to use the same method) and responded to each other’s progress with supportive feedback and suggestions for reaching weight, calorie, and activity goals. Thus, as group sessions became less frequent (weekly sessions, bimonthly sessions, to monthly sessions), intensity of peer support increased. If needed, research staff provided corrective feedback to participants on their correspondence.

Reduced intensity behavioral weight loss program plus mentor support (rBWL+Mentor).

Participants in this arm received the rBWL program described above plus e-support from a mentor. Mentors were individuals who previously completed one of our behavioral weight loss programs (and were, thus, familiar with the intervention) and were successful (i.e. lost at least 10% of their initial body weight and kept it off for at least one year). Mentors were contacted about the present study, and those interested attended an orientation and were consented. Mentors were randomly paired to support two to three participants of the same gender. Unlike in the peer arm, where the provision of support was reciprocal in nature, support in rBWL+Mentor was unidirectional; mentors provided support to participants, but participants did not provide support to mentors. Prior to the first group session, mentors completed a one-hour training in how to provide support using the same strategies noted above in the peer arm.

Mentors attended behavioral weight loss group sessions. In early sessions, mentors and participants engaged in relationship building activities in which they got to know one another, and participants shared how their mentor could help them reach their weight loss goals. Then, identical to the peer support frequency, mentor e-support occurred via email or text (mentors and mentees had to agree on modality) each week that there were no group meetings and, thus, also increased in intensity over time. If needed, research staff provided corrective feedback to participants and mentors on their correspondence. Mentors were not compensated for attending in-person treatment sessions or for providing e-support.

MEASURES.

The primary outcome (weight) was assessed by staff masked to intervention allocation. Participants received a $25 honorarium for completing measures at 3, 6, and 9 months, and a $50 honorarium for completing the 12 month assessment.

Demographics.

At baseline, participants reported sex, age, race, ethnicity, and education.

Weight and height.

Weight was measured to the nearest 0.1kg using a digital scale at every assessment and intervention session. Height was collected at baseline with a wall-mounted stadiometer. BMI was calculated using the formula weight in kg / height in m2. Percent weight loss was calculated as follows: ((baseline weight – current weight)/baseline weight)*100.

Adherence.

Session attendance was objectively recorded by intervention staff at each group meeting. In rBWL+Peer and rBWL+Mentor, compliance with the e-support protocol (sending progress and providing feedback) was objectively measured throughout treatment by having participants “cc” a research assistant (non-intervention staff) on email correspondence and include the research assistant’s email on text correspondence. Communications were reviewed weekly for appropriateness and concerns were addressed immediately.

Social comparison.

Participants in rBWL+Peer and rBWL+Mentor, were asked about social comparison processes with their peer or mentor, respectively, using a validated measure (Leahey & Gokee Larose, 2010; Rancourt, Leahey, LaRose, & Crowther, 2015). Specifically, social comparison direction (upward [participant doing worse than peer/mentor], downward [participant doing better than peer/mentor], or lateral [participant doing same as peer/mentor]) was assessed by having participants indicate how they believe their weight-related progress compared to their peer’s/mentor’s (e.g., Compared to your partner/mentor, how would you rate your current weight loss accomplishments?). Response options range from +2 (“My peer partner/mentor is more successful”; upward comparison) to 0 (“My peer partner/mentor and I are equally successful”; lateral comparison) to −2 (“I am more successful than my peer partner/mentor”; downward comparison). Mean scores were calculated to obtain a social comparison score. To examine social comparison processes over time, participants completed the social comparison measure at week 2, month 3, month 6, month 9, and month 12.

STATISTICAL ANALYSES.

Basic descriptive statistics were conducted to characterize the sample on demographic, weight, and social comparison outcomes. The distribution of percent weight loss and social comparison were examined and adjusted functions were chosen, when necessary, for multilevel and models. Analysis for the main outcome, weight loss, used a multilevel framework such that time was nested within participant. The full model included intervention arm and time defined by treatment phase (Phase I = weekly group sessions [BL-week 6], Phase 2 = bimonthly group sessions [week 7–12], Phase 3 = monthly group sessions [Weeks 13–52]), and the group by treatment phase time interaction. The empty model was obtained to calculate the ICC and random effects on intercepts. The dyadic effects for rBWL+Peer and rBWL+Mentor groups were calculated over time by using ICCs at each time point. Goodness of fit was compared between the empty and full model using four model fit indices: log-likelihood, deviance, Akaike information criteria (AIC), and Bayesian information criteria (BIC) (Burnham, 2011; Raftery, 1995). To determine whether the reduction in residual variance was significant, the empty and full models were compared using the chi-square (χ2) difference test. Missing data were imputed and sensitivity analysis was conducted using multiple imputation by chained equations (Zhang, 2016). Analysis for social comparison were done for rBWL+Peer and rBWL+Mentor.

POWER ANALYSIS.

Estimated mean weight loss trajectories were based on previous 12-month trial results (e.g., Knowler et al., 2002), and variance was assumed to be an increasing function of the mean satisfying σ2=exp(.06*μ), while the correlation matrix was taken as continuous time AR(1) with autocorrelation coefficient ϕ=0.985 per each phase of follow-up. Thus, this trial was powered to detect an effect size of r=.20 at month 12, or about a 1.2% difference in percent weight loss between intervention arms. This is a small-to-medium effect between study arms, as is commonly observed between two active health behavior change interventions and is clinically meaningful (Stevens et al., 1993; Trials of Hypertension Collaborative Research Group, 1997). Power calculations also took into account previous research demonstrating an ICC, or clustering, of weight losses among peer patient dyads and among participants receiving support by the same mentor (Leahey & Wing, 2013). Since power is maximized by equalizing effective sample sizes (N*) accounting for the ICC rather than nominal sample sizes (N) across study groups (Murray, 1998), a sample size of N=272 (rBWL: N=80; rBWL+Peer: N=96; rBWL+Mentor: N=96) was estimated to provide sufficient power to detect an effect for the primary aim. Simulations allowed for 15% attrition at 12 months. These assumptions resulted in 82.2% power to detect an effect for the rBWL+Peer vs. rBWL+Mentor contrast, and 99.4% power to detect an effect for the rBWL+Peer vs. rBWL alone contrast.

RESULTS

PARTICIPANT CHARACTERISTICS.

Participants were 75.9% female, 73.0% non-Hispanic White, and had a mean BMI of 34.7±0.2 kg/m2 (Table 1). There were no significant differences in baseline characteristics by treatment arm (P’s>.11). Retention at trial end was 94.6% (Figure 1).

Table 1.

Baseline participant characteristics.

rBWL
(n=82)
M±SE or % (95% CI)		 rBWL+Peer
(n=98)
M±SE or % (95% CI)		 rBWL+Mentor
(n=98)
M±SE or % (95% CI)
Gender
 Female 76.8 (67.5, 86.2) 73.5 (64.6, 82.5) 77.6 (69.1, 86.0)
Age, y 51.7±0.6 52.1±0.6 51.7±0.5
Race
 African American 9.8 (3.2, 16.3) 5.1 (0.7, 9.5) 10.2 (4.1, 16.3)
 American Indian 1.2 (−1.2, 3.7) 0.0 (0.0, 0.0) 0.0 (0.0, 0.0)
 White 75.6 (66.1, 85.1) 81.6 (73.8, 89.4) 76.5 (68.0, 85.1)
 Multiracial 2.4 (−1.0, 5.9) 2.0 (−0.8, 4.9) 6.1 (1.3, 11.0)
 Other 9.8 (3.2, 16.3) 11.2 (4.9, 17.6) 7.1 (2.0, 12.3)
 Did not report 1.2 (−1.2, 3.7) 0 (0.0, 0.0) 0.0 (0.0, 0.0)
Ethnicity
 Hispanic/Latino 11.0 (4.1, 17.9) 11.2 (4.9, 17.6) 15.3 (8.1, 22.6)
 Non-Hispanic/Latino 87.8 (80.6, 95.0) 88.8 (82.4, 95.1) 84.7 (77.4, 92.0)
 Did not report 1.2 (−1.2, 3.7) 0.0 (0.0, 0.0) 0.0 (0.0, 0.0)
Education
 College graduate 67.1 (56.7, 77.5) 52.0 (42.0, 62.1) 61.2 (51.4, 71.0)
 Non-college graduate 32.9 (22.5, 43.3) 48.0 (37.9, 58.0) 38.8 (29.0, 48.6)
Wt, kg 96.1±1.7 95.1±1.5 95.9±1.3
BMI, kg/m2 34.5±0.4 34.8±0.3 34.8±0.3
Open in a new tab

TREATMENT ADHERENCE.

Percentage of intervention sessions attended did not differ by treatment arm (rBWL=76.4% [70.6, 82.2], rBWL+Peer=79.7% [74.4, 84.9], rBWL+Mentor=82.6% [78.4, 86.7], P=.24). The majority (97.5%) of participants chose to correspond via email and this did not differ by treatment arm (rBWL+Peer: 98.0% corresponded via email; rBWL+Mentor: 96.9% corresponded via email). In rBWL+Peer and rBWL+Mentor, compliance with the e-support protocol was excellent (Figure 2). While the two arms did not significantly differ in overall percentage of weeks in which participants reported their progress to their peer/mentor (rBWL+Peer=84.1% [78.3, 89.9], rBWL+Mentor=78.6% [72.5, 84.7], P=.19), there was a significant group by time interaction such that rBWL+Mentor had a significantly greater decline in progress reports submitted compared to rBWL+Peer (P=.002; Figure 2). There was also a significant difference in percentage of weeks in which peers/mentors responded with feedback, with mentors providing more feedback than peers (rBWL+Peer=90.4% of weeks [87.6, 93.3], rBWL+Mentor=97.7% of weeks [96.8, 98.6], P<.001).

Figure 2.

Figure 2.

Open in a new tab

Percentage of participants in rBWL+Peer and rBWL+Mentor who submitted their progress during each week of coaching, and percentage of participants and mentors in rBWL+Peer and rBWL+Mentor who provided feedback.

WEIGHT LOSS.

There was a significant arm by treatment phase interaction for weight loss (P=.02). While weight losses did not differ among arms during Phases 1 and 2 of treatment (when group meetings were frequent; P’s>.50), in Phase 3, when group sessions became less frequent and peer and mentor e-support became more frequent, rBWL+Peer lost significantly more weight than rBWL alone (P=.001). rBWL+Mentor was not significantly different form the other two arms (P’s>.43). See Figure 3 and Table 2.

Figure 3.

Figure 3.

Open in a new tab

Percent weight loss across treatment phase.

Table 2.

Percent weight loss by treatment phase.

Treatment Phase Arm N Mean SD 95% CI d (Peer/Mentor vs rBWL) d (Mentor vs Peer)
Phase I rBWL 82 −3.61 2.44 −4.17, −3.05
rBWL+Peer 98 −4.14 2.58 −4.71, −3.57 −0.21
rBWL+Mentor 98 −4.01 2.42 −4.51, −3.51 −0.16 −0.05
Phase II rBWL 82 −6.69 4.05 −7.66, −5.72
rBWL+Peer 98 −7.66 4.37 −8.60, −6.71 −0.17
rBWL+Mentor 98 −7.22 3.72 −7.99, −6.45 −0.07 −0.11
Phase III rBWL 82 −7.59 8.13 −9.49, −5.70
rBWL+Peer 98 −8.90 9.22 −10.79, −7.02 −0.16
rBWL+Mentor 98 −8.71 8.22 −10.51, −6.92 −0.13 −0.02
Open in a new tab

Associations between treatment adherence and weight losses were examined. In all three arms, attendance at treatment sessions was associated with better weight loss outcomes (rBWL: r=−.50, P<.001; rBWL+Peer: r=−.49, P<.001; rBWL+Mentor: r=−.56, P<.001). In rBWL+Peer and rBWL+Mentor, participants who submitted more weight management progress to their peer/mentor lost significantly more weight (rBWL+Peer: r=−.41, P<.001, rBWL+Mentor: r=−.48, P<.001). The impact of e-support (providing support and suggestions in response to progress) on weight loss differed by arm. Receiving e-support from a mentor (rBWL+Mentor arm) was not significantly associated with weight loss outcomes (.17, P=.10); in contrast, receiving e-support from a Peer (rBWL+Peer) was significantly associated with weight loss outcomes (r=−.39, P<.001), with more support associated with better weight loss.

SOCIAL COMPARISON & SOCIAL INFLUENCE.

Social comparison direction differed between peer and mentor intervention arms (P<.001). Peer participants had lower social comparison scores at all time points compared to mentor participants (see Table 3). These results suggest that participants in rBWL+Peer engaged in more lateral types of social comparison processes with their peer supporter (partners’ progress is similar), whereas rBWL+Mentor participants engaged in more upward social comparisons with their mentor supporter (mentor’s progress is better). The association between social comparison direction and weight loss was examined by collapsing across treatment arm; results showed that social comparison was significantly associated with weight loss outcomes at all time points (r’s=.30 to .61, P’s≤.01), with upward social comparisons associated with worse weight loss. The associations between social comparison and weight losses were also examined within the peer and mentor treatment arms. Within both rBWL+Peer and rBWL+Mentor, social comparison was significantly associated with weight loss at all time points (rBWL+Peer: r’s=.30 to .42, P’s≤.011; rBWL+Mentor: r’s=.40 to .61, P’s≤.001). Given that rBWL+Peer and rBWL+Mentor did not significantly differ in weight loss outcomes, social comparison was not examined as a mediator of treatment effects.

Table 3.

Social comparison at each assessment time point.

Arm N Mean SD 95% CI d (Mentor vs. Peer)
Week 2 rBWL+Peer 98 −0.05 0.75 −0.22, 0.11
rBWL+Mentor 98 0.37 0.80 0.20, 0.53 0.54
Month 3 rBWL+Peer 98 0.03 0.97 −0.19, 0.25
rBWL+Mentor 98 0.68 0.74 0.53, 0.84 0.75
Month 6 rBWL+Peer 98 0.04 1.01 −0.19, 0.27
rBWL+Mentor 98 0.44 0.79 0.28, 0.60 0.44
Month 9 rBWL+Peer 98 0.13 0.97 −0.09, 0.35
rBWL+Mentor 98 0.34 0.97 0.14, 0.55 0.21
Month 12 rBWL+Peer 98 0.01 1.17 −0.26, 0.28
rBWL_Mentor 98 0.44 0.99 0.24, 0.65 0.40
Open in a new tab

The ICC was examined to explore social influence among peer support partners and among participants with the same mentor. Over the course of treatment, the ICC for weight loss increased among peer dyads in rBWL+Peer (.05 to .20, P=.03). In contrast, in rBWL+Mentor, participants receiving support from the same mentor had similar ICCs at the beginning and end of treatment (.12 to .11, P>.45). This suggests that peer partners tend to influence one another’s weight loss increasingly over time, and consequently achieved somewhat similar weight losses; in contrast, while participants assigned to the same mentor tended to achieve somewhat similar weight losses (ICCs=.12 to .11), mentors’ influence on their mentees was similar across time.

INTERVENTION EFFECTS ON MENTORS.

Prior to the trial, mentors (N=26; 73.1% female, 84.6% non-Hispanic white, age=53.7±1.1, BMI=28.6±0.8) lost an average of 15.3±1.4% of initial body weight and kept it off for at least 1 year. Mentors attended 85.7% of treatment sessions and 97.7% completed the study. During treatment, mentors maintained their weight on average (pre-treatment weight: 79.5±2.8kg, post-treatment weight: 80.7±2.6kg, P=.15), with the majority of mentors (69.2%) losing additional weight or maintaining their weight loss in full. A minority (30.8%) regained weight (defined as gaining 2.3kg or more (Wing, Tate, Gorin, Raynor, & Fava, 2006)).

DISCUSSION

The INSPIRE trial examined the efficacy of peer vs. mentor e-support when added to a reduced intensity obesity treatment, and explored social comparison processes within peer dyads and between mentors and mentees. Results showed that adding between-session peer support to a reduced intensity behavioral weight loss program improved weight loss outcomes when intervention sessions became infrequent. In contrast, adding mentors to the reduced intensity program did not significantly improve weight losses and there was no significant difference in weight losses between peers and mentors. As hypothesized, social comparison processes differed between peer participants and mentor participants, with peer dyads reporting more lateral social comparisons and mentees reporting more upward social comparisons with their mentors. Further, upward social comparisons were linked to poorer weight loss outcomes.

The weight loss results are clinically meaningful and are largely consistent with the peer and mentor support literature. In the present study, differences in weight loss between the rBWL+Peer arm vs. rBWL alone arm were of sufficient magnitude to produce improvements in cardiovascular disease risk factors (Trials of Hypertension Prevention Collaborative Research Group, 1997; Stevens et al., 1993). Results are also consistent with the peer support literature. In other studies, peers (two new patients actively supporting one another) have been shown to improve health behavior and outcomes. Consistent with the current findings, Heisler and colleagues found that adding peer support to active diabetes treatment resulted in significant improvement in glycemic control (Heisler et al., 2010). The effects of mentor support (previous, successful patient helping new patients) on health behaviors and outcomes has been more mixed. While mentors have been shown to improve cardiovascular disease risk factors when added to an educational control (Tang, Funnell, Sinco, Spencer, & Heisler, 2015), adding mentors to an active diabetes self-management program had no effect on HbA1c (Keyserling et al., 2002). While there were no previous fully powered trials with a head-to-head comparison of peer vs. mentor patient support, our pilot study showed that peer support yields significantly better weight loss outcomes compared to mentor support (Leahey & Wing, 2013). Results from the current trial and those of previous studies suggest that peers may be more effective than mentors when added to an active treatment (vs. no treatment / educational control). It is possible that because peers initiate weight loss at the same time, and thus actively face the same health behavior change challenges, they may provide one another with a unique sense of empathy that isn’t possible from very successful mentors. Further, it’s worth noting that, compared to mentor participants, peers had more sustained compliance with submitting their progress to one another over time in the present study. This may be an indication that peer dyads, who relied on one another for support, felt more responsible to each other during treatment. Additional research is needed that contributes to our understanding as to why these two types of patient support may differentially affect outcomes by exploring mechanisms of change that differ between the two types of patient supporters.

Social comparison may be one mechanism by which peers and mentors yield differential treatment effects. Peers are naturally lateral comparison targets (two new patients just starting behavior change are “similar” to one another) whereas mentors are naturally upward comparison targets (new patient is by nature “worse off” than successful mentor). Results herein support this assertion. Peer dyads reported social comparison scores consistent with lateral comparisons, whereas mentees reported social comparison scores consistent with upward comparisons. This suggests that, on average, peers’ and mentors’ level of success is quite salient such that participants’ social comparison perceptions of success are consistent with actual success. Interestingly, while upward comparisons were associated with worse weight loss outcomes in the present study, there was no significant difference between peer and mentor arms on weight loss. It’s possible that upward social comparison processes in the mentor arm may have undermined the effects of mentor support but not of sufficient degree to cause differential effects on treatment outcomes between peers and mentors. Additional research is needed to better understand whether and why differences may exist between mentor and peer support. For example, our finding that peer feedback was associated with weight loss outcomes but mentor feedback was not suggests that investigating the content of peer vs. mentor support/feedback may elucidate mechanisms of action that differ between these two types of supporters. Plausible mechanisms include various types of social support (e.g., directive support, autonomy support) and social accountability.

Interestingly, our ICC results suggest social influence between peer supporters and mentors and mentees. Weight losses among peer dyads tended to become more similar over time, such that at the end of treatment, the ICC in weight loss among peers was consistent with a large effect. Mentees coached by the same mentor also tended to achieve similar weight losses, but the effect was consistent throughout treatment and was relatively smaller. These results are similar to those of previous studies demonstrating that peer support dyads achieve similar HbA1c outcomes (Heisler et al., 2010). As noted previously, future studies may further examine the social influence processes that occur between peer dyads / mentors and mentees to maximize the effects of peer and mentor support on health outcomes.

The present study examined the impact of being a mentor on mentors’ weight trajectory. Nearly 70% of mentors involved in this study maintained their weight loss in full or lost additional weight. This is consistent with previous findings in the substance abuse literature suggesting that being a sponsor / mentor may be beneficial to the sponsor (Crape et al., 2002). While it is impossible to know whether these individuals would have been successful regardless of being a mentor, it may be worth exploring whether becoming a mentor after initial weight loss treatment leads to better, sustained weight loss maintenance.

This study has some limitations. The sample was predominantly female and non-Hispanic White, which limits generalizability of study results to men and racial/ethnic minorities. In addition, because this study was an efficacy trial, it included a behavioral run–in; thus, results may not generalize to non-research programs that do not include a run-in. Related to the efficacy design, this study did not report on the cost-effectiveness of adding peers or mentors to the otherwise reduced intensity program. Such information would be useful from a pragmatic perspective to determine whether adding patient support is worth the increased cost (Glasgow et al., 2012).

This study has notable strengths. The three arm randomized controlled trial design allowed for an experimental manipulation to determine whether peer or mentor support improves weight loss outcomes when added to a reduced intensity program. Further, the study was informed by Social Comparison Theory (Festinger, 1954); thus, findings will both enrich theory and contribute to our understanding of how social comparison processes work in the context of peer and mentor health behavior change paradigms. Moreover, a reduced intensity behavioral weight loss program was included herein that is consistent with what is being offered in clinical care. Thus, our finding that the addition of peer e-support may be particularly effective for improving longer-term weight management in otherwise reduced intensity interventions has potential for real-world translation. Further, given that peers (two patients actively working on health behavior change) may be more easy to identify and recruit than mentors (i.e. extremely successful patients), utilizing peers in clinical practice may be more pragmatic, efficient, and effective. Taken together, in light of the use of reduced intensity programs by the CMS, our findings suggesting that the addition of peer support may improve outcomes in such programs and provide an evidence base to spur public policy to promote reimbursement of peer support for obesity treatment. These findings also have implications for weight loss maintenance; given evidence that continuous care is most effective for maintenance (Perri et al., 1987), peer e-support may be a particularly effective strategy for providing such care in an organically sustainable, cost-effective manner. Finally, the primary outcome (weight) was objectively assessed, and retention and adherence rates were excellent, allowing for a robust examination of the efficacy of peer and mentor support on obesity treatment outcomes.

Supplementary Material

Supplemental Material

Acknowledgments

This trial was supported by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health (NIH) under award number R01DK095771. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. This trial is registered at clinicaltrials.gov (NCT01642199).

REFERENCES

  1. Aspinwall LG, & Taylor SE (1993). Effects of social comparison direction, threat, and self-esteem on affect, self-evaluation, and expected success. Journal of Personality and Socical Psychology, 64(5), 708–722. [DOI] [PubMed] [Google Scholar]
  2. Burnham KPA, D. R.; Huyvaert KP. (2011). AIC model selection and multimodel inference in behavioral ecology. Behavioral Ecology and Sociobiology, 65, 23–35. [Google Scholar]
  3. Buunk J, & Ybema J (1997). Social comparison and occupational stress: The identification contrast model. In Buunk FXGBP (Ed.), Social comparison, health, and coping (pp. 359–388). Mahwah: Erlbaum. [Google Scholar]
  4. Carvajal R, Wadden TA, Tsai AG, Peck K, & Moran CH (2013). Managing obesity in primary care practice: a narrative review. Annals of the New York Academy of Sciences, 1281, 191–206. doi: 10.1111/nyas.12004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Castro CM, Pruitt LA, Buman MP, & King AC (2011). Physical activity program delivery by professionals versus volunteers: The TEAM randomized trial. Health Psychology, 30(3), 285–294. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Chu DT, Nguyet NTM, Dinh TC, Lien V,T, Nguyen K, & Ngoc VT,N et al. (2018). An update on phsycial health and economic consequences of overweight and obesity. Diabetes & Metabolic Syndrome: Clinical Research & Reviews, 12(6), 2095–1100. [DOI] [PubMed] [Google Scholar]
  7. Centers for Medicare and Medicaid Services (2011). Proposed Decision Memo for Intensive Behavioral Therapy for Obesity (CAG-00423N). Retrieved from http://www.cms.gov/medicare-coverage-database/details/nca-proposed-decision-memo.aspx?NCAId=253&ver=2&NcaName=Intensive+Behavioral+Therapy+for+Obesity&TimeFrame=7&DocType=All&bc=AgAAYAAAIAAA&
  8. Collins R (1996). For better or worse: The impact of upward social comparison on self-evaluations. Psychological Bulletin, 119, 51–69. [Google Scholar]
  9. Crape BL, Latkin CA, Laris AS, & Knowlton AR (2002). The effects of sponsorship in 12-step treatment of injection drug users. Drug and Alcohol Dependence, 65(3), 291–301. [DOI] [PubMed] [Google Scholar]
  10. Dennis CL (2003). Peer support within a health care context: a concept analysis. International Journal of Nursing Studies, 40(3), 321–332. [DOI] [PubMed] [Google Scholar]
  11. Diabetes Prevention Program Research Group (2002). The Diabetes Prevention Program : Description of the Lifefstyle Intervention. Diabetes Care, 25, 2165–2171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dorgo S, Robinson KM, & Bader J (2009). The effectiveness of a peer-mentored older adult fitness program on perceived physical, mental, and social function. Journal of the American Academy of Nurse Practioners, 21(2), 116–122. [DOI] [PubMed] [Google Scholar]
  13. Festinger L (1954). A theory of social comparison processes. Human Relations, 7, 117–140. [Google Scholar]
  14. Finkelstein EA, Khavjou OA, Thompson H, Trogdon JG, Pan L, Sherry B, & Dietz W (2012). Obesity and severe obesity forecasts through 2030. American Journal of Preventive Medicine, 42(6), 563–570. doi: 10.1016/j.amepre.2011.10.026 [DOI] [PubMed] [Google Scholar]
  15. Fitzpatrick SL, Wischenka D, Appelnahs BM, Pbert O, Wang M, Wilson DK, Pagoto SL, Society of Behavioral Medicine (2016). An evidence-based guide for obesity treatment in primary care. The American Journal of Medicine, 115, e1–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Glasgow RE, Gaglio B, Bennett G, Jerome GJ, Yeh HC, Sarwer DB, Appel L, et al. (2012). Applying the PRECIS criteria to describe three effectiveness trials of weight loss in obese patients with comorbid conditions. Health Services Research, 47, 1051–1067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Hales CM, Carroll MD, Fryar CD, & Ogden CL Prevalence of obesity among adults and youth: United States, 2015–2016. National Center for Health Statistics Data Brief, no. 288 Hyattsville, MD: National Center for Health Statistics, 2017. [Google Scholar]
  18. Heisler M, & Piette JD (2005). “I help you, and you help me”: facilitated telephone peer support among patients with diabetes. Diabetes Educator, 31(6), 869–879. [DOI] [PubMed] [Google Scholar]
  19. Heisler M, Vijan S, Makki F, & Piette JD (2010). Diabetes control with reciprocal peer support versus nurse care management: a randomized trial. Annals of Internal Medicine, 153(8), 507–515. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hibbard MR, Cantor J, Charatz H, Rosenthal R, Ashman T, Gundersen N, … Gartner A (2002). Peer support in the community: initial findings of a mentoring program for individuals with traumatic brain injury and their families. Journal of Head Trauma Rehabilitation, 17(2), 112–131. [DOI] [PubMed] [Google Scholar]
  21. Keyserling TC, Samuel-Hodge CD, Ammerman AS, Ainsworth BE, Henriquez-Roldan CF, Elasy TA, … Bangdiwala SI (2002). A randomized trial of an intervention to improve self-care behaviors of African-American women with type 2 diabetes: impact on physical activity. Diabetes Care, 25(9), 1576–1583. [DOI] [PubMed] [Google Scholar]
  22. Knowler WC, Barrett-Connor E, Fowler SE, Hamman RF, Lachin JM, Walker EA, … Diabetes Prevention Program Research, G. (2002). Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. New England Journal of Medicine, 346(6), 393–403. doi: 10.1056/NEJMoa012512 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Leahey T, Gokee Larose J. (2010). Social comparisons among members of a behavioral weight loss program. Obesity, 18, S108. [Google Scholar]
  24. Leahey TM, Crowther JH, & Mickelson KD (2007). The frequency, nature, and effects of naturally occurring appearance-focused social comparisons. Behavior Therapy, 38(2), 132–143. [DOI] [PubMed] [Google Scholar]
  25. Leahey TM, & Wing RR (2013). A randomized controlled pilot study testing three types of health coaches for obesity treatment: Professional, peer, and mentor. Obesity (Silver Spring), 21(5), 928–934. doi: 10.1002/oby.20271 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Lockwood P, & Kunda Z (1997). Superstars and me: Predicting the impact of role models on the self. Journal of Personality and Social Psychology, 73, 91–103. [Google Scholar]
  27. Murray D (1998). Design and analysis of group-randomized trials. New York: Oxford University Press. [Google Scholar]
  28. NIH. (1998). Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults--The Evidence Report. National Institutes of Health. Obesity Research, 6 Suppl 2, 51S–209S. [PubMed] [Google Scholar]
  29. Olshansky SJ, Passaro DJ, Hershow RC, Layden J, Carnes BA, Brody J, … Ludwig DS (2005). A potential decline in life expectancy in the United States in the 21st century. New England Journal of Medicine, 352(11), 1138–1145. doi: 10.1056/NEJMsr043743 [DOI] [PubMed] [Google Scholar]
  30. Parent N, & Fortin F (2000). A randomized, controlled trial of vicarious experience through peer support for male first-time cardiac surgery patients: impact on anxiety, self-efficacy expectation, and self-reported activity. Heart Lung, 29(6), 389–400. [DOI] [PubMed] [Google Scholar]
  31. Perri MG, McAdoo WG, McAllister DA, Lauer JB, Jordan RC, Yancey DZ, & Nezu AM (1987). Effects of peer support and therapist contact on long-term weight loss. Journal of Consulting and Clinical Psychology, 55(4), 615–617. [DOI] [PubMed] [Google Scholar]
  32. Raftery A (1995). Bayesian model selection in social research. Social Metholdology, 25, 111–163. [Google Scholar]
  33. Rancourt D, Leahey TM, LaRose JG, & Crowther JH (2015). Effects of weight-focused social comparisons on diet and activity outcomes in overweight and obese young women. Obesity (Silver Spring), 23(1), 85–89. doi: 10.1002/oby.20953 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Riegel B, & Carlson B (2004). Is individual peer support a promising intervention for persons with heart failure? Journal of Cardiovascular Nursing, 19(3), 174–183. [DOI] [PubMed] [Google Scholar]
  35. Stevens FJ, Corrigan SA, Obarzanek E, et al. (1993). WEight loss intervention in phase 1 of the Trials of Hypertension Prevention. Archives of Internal Medicine, 153, 849–858. [PubMed] [Google Scholar]
  36. Stewart ST, Cutler DM, & Rosen AB (2009). Forecasting the effects of obesity and smoking on U.S. life expectancy. New England Journal of Medicine, 361, 2252–2260. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Tang TS, Funnell MM, Sinco B, Spencer MS, & Heisler M (2015). Peer-Led, Empowerment-Based Approach to Self-Management Efforts in Diabetes (PLEASED): A Randomized Controlled Trial in an African American Community. Annals of Family Medicine, 13 Suppl 1, S27–35. doi: 10.1370/afm.1819 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Tesser A, Millar M, & Moore J (1988). Some affective consequences of social comparison and reflection processes: the pain and pleasure of being close. Journal of Personality and Social Psychology, 54(1), 49–61. [DOI] [PubMed] [Google Scholar]
  39. The Trials of Hypertension Prevention Collaborative Research Group (1997). Effects of weight loss and sodium reduction intervetion on blood pressure and hypertension incidence in overweight peole with high-normal blood pressure. Archives of Internal Medicine, 157, 657–667. [PubMed] [Google Scholar]
  40. Thomas S, Reading J, Shephard RJ (1992). Revision of the Physical Activity Readiness Questionnaire (PAR-Q). Canadian Journal of Sports Sciences, 17, 338–335. [PubMed] [Google Scholar]
  41. Tsai AG, Remmbert JE, Butryn ML, & Wadden TA (2018). Treatment of obesity in primary care. Medical Clinics of North America, 102, 35–47. [DOI] [PubMed] [Google Scholar]
  42. Wing RR, Tate DF, Gorin AA, Raynor HA, & Fava JL (2006). A self-regulation program for maintenance of weight loss. New England Journal of Medicine, 355(15), 1563–1571. doi: 10.1056/NEJMoa061883 [DOI] [PubMed] [Google Scholar]
  43. Zhang Z (2016). Multiple imputation with multivariate imputation by chained equation (MICE) package. Annals of Translational Medicine, 4(2), 30. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Material
NIHMS1727594-supplement-Supplemental_Material.pdf (182.7KB, pdf)

RESOURCES