Comparing the Effects of Teen Mentors to Adult Teachers on Child Lifestyle Behaviors and Health Outcomes in Appalachia

Abstract

Childhood obesity prevalence rates in the United States are the highest in the rural Appalachian areas. Teens mentoring younger children to reverse obesity health risks are an understudied approach. This randomized-controlled trial compared the effects of two curriculum delivery methods and assessed the mediating effects of the number of sessions attended on the outcomes. The control group received the 8-week Just for Kids! curriculum via an adult teacher in a classroom and the experimental group received the same curriculum via individual teen mentoring. Data collected at baseline and postintervention were analyzed using multilevel linear models. Each of the outcomes (e.g., body mass index, blood pressure, current lifestyle behaviors) were modeled separately. Only the mentored children demonstrated improved current lifestyle behaviors (e.g., physical activity and dietary patterns) and health outcomes. Teen mentoring was an effective and efficacious approach to impact the lifestyle patterns and health outcomes of children in a school setting.

Childhood obesity is a national epidemic. The proportion of third and fourth grade children in the United States who meet criteria for obesity is fast approaching 25% (Montgomery-Reagan, Bianco, Heh, Rettos, & Huston, 2010). Traditionally, rural areas have experienced lower prevalence of childhood overweight and obesity due to the increased physical demands characteristic of an agrarian lifestyle. However, this no longer is the case; rural children now show an increased prevalence of obesity and overweight compared to their urban counter-parts (Montgomery-Reagan et al., 2010). Estimates for rural-dwelling children exceed the national averages by as much as double (Bellamy, Bolin & Gamm, 2011; Oza-Frank, Norton, Scarpitti, Wapner, & Conrey, 2011). Furthermore, in rural areas across the United States, childhood overweight and obesity prevalence rates are highest in rural Appalachia (Bellamy, Bolin & Gamm, 2011; Centers for Disease Control and Prevention, 2008; Montgomery-Reagan et al. 2010; Smith & Holloman, 2011). Appalachian children have overweight and obesity prevalence rates approaching 40%; this rate exceeds all other racial, demographic, or geographical groups (Daniels, 2009, Oza-Frank et al., 2011).

Childhood obesity is highly associated with obesity in adulthood, which has well-known comorbidities. The problem is particularly severe among lower income and rural populations (Montgomery-Reagan et al., 2010) who also have elevated rates of diseases related to diet and physical activity, such as type 2 diabetes, hypertension, and cardiovascular disease (Daniels, 2009). This is a major concern in rural Appalachian populations that have relatively poor health and health-related self-care behaviors compared to other populations (Behringer & Friedell, 2006; Tai-Seale & Chandler, 2003).

Rural life presents special challenges to maintaining a healthy weight, including structural and cultural factors that prevail in Appalachian areas (Bellamy, Bolin & Gamm, 2011; Tai-Seale & Chandler, 2003). Cultural factors include higher fat and caloric consumption, less exercise, preference for sedentary activities, reliance on nonprofessional health advice, preference for informal communication channels, and less confidence in the recommendations of teachers or health professionals (Bellamy, Bolin & Gamm, 2011; Janicki et al., 2011; Tai-Seale & Chandler, 2003). Finally, rural Appalachian residents lack nutritional education and adequate resources for either healthful eating or exercise (Janicki et al., 2009). However, effective interventions to improve lifestyle patterns, specifically dietary and activity patterns are sorely understudied in rural Appalachian populations.

In targeting children, mentoring approaches have been effective at addressing risk behaviors (Karcher, 2008; Smith & Barker, 2009). Mentoring is a dyadic psychosocial intervention wherein a more experienced or knowledgeable individual is brought into a close relationship with a less knowledgeable person in order to provide support and guidance (Karcher, Nakkula, & Harris, 2005). The presence of a mentor in the life of a young person supports healthy growth and development and serves as a protective buffer against many risks faced by today’s children (Portwood, Ayers, Kinnison, Waris, & Wise, 2005). Mentors help children overcome personal and social barriers, expose them to new relationships and opportunities, and assist in developing decision-making and problem-solving skills (Portwood et al., 2005). Mentoring relationships have positively influenced behavior change and health outcomes including academic achievement (Karcher, 2008), substance use (Aseltine, Dupree, & Lamlein, 2000), and lifestyle behaviors (Smith, 2011). However, the use of mentoring has not been adequately tested. Most published studies were non-experimental, focused only on satisfaction levels as outcomes (Hanson, 2003; Sawyer, 2001), and did not distinguish between adult and teen mentors (Sheehan, DiCara, LeBailly, & Christoffel, 1999).

Particularly puzzling is that the use of older teens to mentor younger children has not received more attention. By infusing local residents as health coaches, teen mentoring to reverse health risks and promote healthy lifestyle behaviors may impact the rising problem of childhood obesity in rural Appalachian and other more remote or underserved populations. Using teens as mentors has been equally effective compared to adult mentors in promoting school connectedness in young children (Karcher, Davis, & Powell, 2002). Sheehan, DiCara, LeBailly, and Christoffel (1999) found a positive effect on cross-age mentee’s classroom behavior and attitudes toward risk behaviors. Finally, Westerman (2002) found that in a rural Appalachian area, teen mentors helped younger mentees achieve improved academic achievement.

In the teen-mentoring approach, linkages are created through the mentor-mentee dyads. The linkages provide children with emotional, informational and appraisal support that creates a sense of psychological safety (between the mentor and mentee) resulting in more positive attitudes toward changing behavior (Smith & Barker, 2009). Learning, self-efficacy, and behavior change are facilitated when people have a sense of psychological safety or the perception that attempts to change behavior can occur without fear or embarrassment (Heaney & Israel, 2008). Consequently, mentoring facilitates psychological safety and subsequently learning. The mentoring approach’s advantage over traditional teacher-led classroom settings is the enhanced learning and support for behavioral change that results from the perceived social support and psychological safety that is promoted by mentoring. Consequently, mentoring provides children with increased control over lifestyle behaviors that can be difficult to change and influenced by a multitude of factors.

Research also has suggested that by providing supportive contexts, older siblings regularly serve as natural mentors and make considerable contributions to their younger siblings’ social and cognitive development (Heaney & Israel, 2008). By modeling empathy and perspective taking, older siblings provide younger siblings opportunities to develop their own empathetic and perspective-taking skills (Karcher, 2005). According to Harris (1998), older peers also are enormously powerful influencing agents. Harris argues that compared to older peers, adults have less influence on younger children because from the child’s point of view, the goal of development is “wanting to be like the bigger kid” … It is the child’s equating of maturity with status that makes younger children want to behave like older peers (p. 267).

Although many mentoring programs are community based, in rural settings, schools are an obvious setting for the implementation of mentoring (Karcher, Davis, & Powell, 2002). To broaden the range of processes and outcomes affected, many school-based programs are beginning to infuse mentoring into their activities (Roth & Brooks-Gunn, 2003). For example, inclusion of mentoring to school curricula has led to child improvement in skill development, self-definition, and sense of self-worth (Kuperminic et al., 2005).

School-based mentoring programs benefit not only the child and mentor but also the school by establishing social networks into the school, thus enhancing connectedness to one’s school and community (Portwood et al., 2005). It has been suggested that school connectedness is the single most effective aspect of the school environment in predicting healthier behaviors and better health (Kuperminic et al., 2005). Based on what is known about successes of adult mentoring and promising results from earlier studies of teens as mentors, this study builds on previous studies. The use a rigorous design that compares teen mentors to adult teachers in delivering health curricula addresses a significant gap in the mentoring literature.

The purpose of this study was to expand the use and evaluation of teen mentoring. Through a school-based randomized-controlled trial, a more rigorous examination of behavioral outcomes and health outcomes builds on previous studies and delivery settings. Our aim was to determine the effectiveness of a health education curriculum for third and fourth grade children delivered by trained teen mentors compared to the same content delivered in a classroom setting by an adult teacher (usual format) in an after school setting. To better understand the dose effect of the curriculum on outcomes, the number of sessions attended by every child was evaluated. The hypotheses were as follows.

Compared to the teacher-led group, at postintervention children in the teen-mentored group will:

1. report a greater increase in current lifestyle behaviors;

2. demonstrate greater improvement in health outcomes (a) body mass index (BMI) for age and gender and (b) blood pressure;

3. report a greater increase in outcome mediators (a) nutritional knowledge; (b) positive attitudes toward eating healthfully; (c) positive attitudes toward engaging in physical activity; (d) intentions; (e) perceived support; and (f) self-efficacy;

4. within each group, children attending a greater number of sessions will show more improvement in each outcome.

Method

This study was a randomized-controlled trial that compared the effects of two curriculum delivery methods and assessed the mediating effects of the number of sessions attended on the outcome measures. The comparison group received the Just for Kids! curriculum via an adult classroom teacher (usual format) and the experimental group received the same curriculum via individual teen mentoring. Just for Kids! was designed for third and fourth graders and modeled after the “Shapedown” program developed at the University of California-San Francisco (Johnson & Mellon, 2003). It addresses the roles of exercise and food in promoting health, moderation in sedentary activities, and encourages children to set reasonable behavioral goals. Additional behaviors targeted include eating regular meals and eating in response to hunger and satiety. It also addresses self-acceptance, processing emotions, assertiveness, positive attitudes, and efficacy. Consistent with the focus of the Just for Kids! the effectiveness of the health education curriculum was measured using concepts from the theory of planned behavior, self-determination theory, and social cognitive theory (see Figure 1). The study was approved by the Social and Behavioral Human Subjects Review Board at the host institution.

Figure 1.

Effects of the health curriculum delivery on the mediators and outcomes.

Population, Recruitment, and Sample

The target population for this study consisted of third and fourth grade students in a predominately poor rural county of Appalachia. They are an understudied aggregate with disproportionately higher rates of obesity and diabetes compared to other regions in the United States (Kumanyika & Grier, 2006; Oza-Frank et al., 2011). In the targeted community, over 31% of third graders are obese and 47% are classified as overweight, while nationwide 17% of adolescents and children are classified as obese with adolescent rates exceeding younger children (Ogden, Carroll, Kit, & Flegal, 2012). Schools were selected based on similar demographic and academic profiles of students, similar student populations, and proximity to local high schools. Based on these criteria, four elementary schools and two high schools participated.

Recruitment occurred at the start of the school year. All third and fourth graders were eligible as long as they had transportation home. Tenth and 11th graders were recruited to serve as cross-age mentors if they (a) had interest in working with younger children, (b) wanted to cultivate their own health-supportive behaviors, (c) were are not expected to move from the school before the conclusion of the study; and (d) earned predominately As and Bs in most classes. To comply with a funding agency requirement, teens with a BMI (for age and gender) that exceeded the 85th percentile at the start of the study were excluded. Adult teachers were recruited through the school. Adult teachers were selected based on the recommendation/recommendations of the After School Network and/or host school.

A brief presentation was given during an assembly to all third and fourth graders during a designated school day. During the assembly, the project staff was present, answered questions from possible participants, and distributed information packets to be taken home. Separate presentations to the targeted child participants and teen mentors were conducted. The information packets contained a letter to parents that described the project and written instructions for the completion and return of consent forms. During the presentation, the study purpose, expected activities, duration, incentives, consent process, and assent process were discussed. Time to answer any questions was provided. Child assent and parental consent were obtained for all participants. Study recruitment lasted 2 weeks. As specified in approved consent and assent forms and to aid study retention, child participants, including the control group received a $25.00 incentive with $10.00 given at baseline data collection and $15.00 given at the postintervention data collection.

All teen mentors attended the same school district as the child participants. Participation as a teen mentor was promoted to all eligible teens however, teens identified as potential mentors by the County After-School Network were given first choice. In addition, information sheets describing the project and the mentoring duties as well as incentives were provided to the participating high schools and distributed to other interested teens. Participating teen mentors received a $100.00 incentive distributed as follows: $20.00 at completion of mentors’ training, $5.00 at completion of each intervention session, and $40.00 after the return of study materials and supplies. Furthermore, teen mentors were given a certificate of leadership achievement and provided with a letter of reference describing the teen’s contributions to the project.

Adult teachers were recruited through the County After School Network. The adult teachers were selected based on the recommendation/recommendations of the After School Network and/or host school and the needs of the study such as matching gender with available child participants. Inclusion and exclusion criteria for the adult teachers were (a) meet the hiring qualifications of the local school district and the State of Ohio, (b) be able to speak and read English, (c) not expected to move from the participating school district/districts before the conclusion of the study, and (d) will reside in the same regional cultural area as the participating schools. Participating adult teachers were paid a total of $500.00 with $50.00 given at the completion of teachers’ training, $50.00 at the completion of each intervention session, and $50.00 after the return of study materials and supplies.

A total of 160 elementary-age children, 32 teen mentors, and 5 adult teachers participated. The child participants were from 3 public schools in the same county. The teen mentors were from the local high schools affiliated with the participating elementary schools. The adult teachers were school employees of the elementary schools such as librarians, secretarial support, and program assistants in the after school programming. Consistent with the surrounding community, most child participants were Caucasian (93.1%) and non-Hispanic (98.1%). Slightly more males (52%) and third graders (56.6%) participated. See Table 1 for additional sample demographics.

Table 1.

Demographic Description of Child Sample (N = 160).

Variable of Interest	Percentage	N
Gender
Male	52	83
Female	48	77
Race
Caucasian/White	93.1	148
African American/Black	3.8	6
Asian American	0.6	1
Other	1.9	3
Hispanic or Latino
Yes	1.9	3
No	98.1	155
Grade in school
Third	56.6	90
Fourth	43.4	69
School attending
School A	41.9	67
School B	21.3	34
School C	36.9	59
Lives with (biological) mother
Yes	86.7	137
No	13.3	21
Lives with (biological) father
Yes	55.1	87
No	44.9	71
Body mass index classification at Pretest
Underweight	0.7	1
Normal weight	51.4	75
Overweight	18.9	28
Obese	29.7	44
Assigned group
Intervention (cross-age mentor)	45.0	72
Control (adult leader)	55.0	88
Mean age
9.24 years (SD = .68)	Range = 8.0–11.08 years

Intervention

For this study, the mentored and adult-led groups met immediately after school for 60 min during the same day of the week but in separate building areas. Each 60-min session consisted of 45 min of structured activities and 15 min of noncompetitive physical activities. Each school hosted the program on a different day of the week. The teen mentors and their assigned mentees met in the gymnasium where other mentor-mentee dyads were present. To minimize distractions, mentoring maintained 1:1 ratios. The mentoring dyads also were distanced from other dyads as much as feasible. Outside distractions were monitored and controlled via limiting access to the room during the sessions. To control for seasonal effects, the intervention was delivered during the same months at all schools.

To maintain curricular integrity, each teen mentor and adult teacher was trained in the delivery of the curriculum and be provided with an Instructor’s Guide that contained (a) the overview and purpose of the program, (b) weekly lesson plans, (c) cues and prompts to deliver the content, (d) structured activities, (e) needed materials for the weekly activities, (f) content summaries to end each session, and (g) copies of all forms and handouts. Curricular workbooks developed for Just for Kids were provided to each child participant which contained homework, worksheets, themed stories, and take-home (family) activities. Weekly topics included keeping your body healthy, the importance of exercise, food groups and the My Pyramid, portion control, emotional eating, food cravings, and building more activity into your daily life.

To ensure program and study integrity, teen mentors and adult teachers met weekly (in separate sessions) with the project director for debriefing. During these 30-min debriefing sessions, the project director assessed consistency of messages delivered; reinforced follow-up messages to be delivered; provided prompts and preparation for the following week; and troubleshot any concerns. The project director supervised teen and child interactions as well as adult and child interactions during the weekly curricular sessions.

Data Collection

A research assistant blinded to group assignment collected all data in a designated private room at each participating elementary school. The research assistant read questions aloud to compensate for any reading difficulties among children and was available to answer any questions. Two adaptations were made in light of the young age of the children. First, children’s data were collected over 2 days for each time point to minimize response burden. At each collection time point, half responded first to physical activity questions and the other half of the children responded first to dietary questions. The remaining halves of data were collected during the second day. Second, the scales used a 5-point response format rather than the usual 7-point format which is more cognitively complex.

Measures

Scales are comprised of items adapted from published studies showing acceptable psychometrics ranging from α of .70–.95 and validity when used with children (Armitage, 2005; Bebestos, Chronis, & Theodorakis, 2002; Hagger, Chatzigarantis, Culverhouse, & Biddle, 2003). Internal consistency reliability coefficients for this study ranged from α of 70–82. Child participants completed each survey measure in approximately 20 min. Height and weight measurements were completed in approximately 3 min per child. Blood pressure measurements were completed in approximately 5 min per child.

Demographic data were collected from the parents during the consent/assent process. Variables of interest included gender, age, grade in school, race or ethnicity, and name of school (see Table 1). The number of sessions attended was recorded for all children.

Health status was measured by BMI for age and blood pressure. BMI was calculated for each child and possible teen mentor. BMI measurement calculations were obtained using the formula: Weight (lb)/Height (in)² × 703. Using CDC guidelines for BMI by age and gender, healthy weight was defined as between the 25th and 85th percentile; overweight was defined as between the 85th and 95th percentile; and obese was defined as above the 95th percentile (Krebs et al., 2007).

Blood pressure readings were obtained after a 15-min resting period with children seated, feet resting flat on a surface, and the right arm resting at heart level (U.S. Department of Health and Human Services, 2005). Standard protocol for children younger than 12 years of age was used. Blood pressures were obtained by rapidly inflating to the maximum inflation level and deflating at a rate of 2 mmHg per second, with 30–60 s between blood pressure determinations.

Current lifestyle behaviors were measured by two sub-scales developed by the researchers and based on the goals of the Just for Kids curriculum (James & Connelly, 2007). Content and concurrent validity were assessed prior to administration of the subscales. The first subscale consisted of five questions and asked about dietary behaviors. The second subscale consisted of 3 items that asked about physical activity behaviors. For each prompt children indicated their level of agreement using a 5-point Likert-type scale with 5 = strongly agree and 1 = strongly disagree. The internal consistency reliability coefficient was α = .73. “Free” and “light” foods, defined for the children as foods low in sugar and fat, were asked about using the prompt: “In the past week I ate … ” (a) at least one fruit a day, (b) free or light foods at lunch each day, (c) only free or light foods at dinner each day, (d) only free or light foods 5 or more days a week, (e) breakfast each day; and “did an activity on my own time, in addition to what I did in school that made …” (e) my heart beat faster, (f) me out of breathe, and (g) me sweat.

Mediators

Nutritional knowledge, attitudes, perceived support, self-efficacy, and intention known to influence or mediate dietary behaviors and physical activity. A 15-item nutritional questionnaire that asked respondents to circle the healthier of two matched choices measured knowledge. Correct responses were summed for a total correct score. The questionnaire was the nutritional knowledge quiz developed for and provided as part of the Just for Kids curricula.

Attitudes toward eating healthfully were measured by having children respond to the prompt “I think for me, eating healthfully in the next week would be … ” with 10 semantically differential items (“bad/good,” “fun/boring”), scored from 1 to 5, with 5 = strongly agree and 1 = strongly disagree. The internal consistency reliability coefficient was α = .80. Children’s attitudes toward eating have been found to have predictive validity regarding dietary behaviors (Conner, Norman, & Bell, 2002). Attitudes toward being physically active were measured with the same prompt used for eating healthfully with 10 semantically differential items (“bad/good,” “fun/boring”) scored from 1 to 5, with 5 = strongly agree and 1 = strongly disagree. The internal consistency reliability coefficient was α = .82 and predictive validity has been established for physical activity behaviors (Conner et al., 2002).

Perceived support was measured by a 6-item Likert-type scale, two sets of 3 items taken from the scale used by Chatzisarantis, Hagger, Biddle, and Smith (2005) for the behaviors of eating healthfully and engaging in physical activity. One set of 6 items asked all children to rate support elements from “People important to me,” ranging from 1 = strongly disagree to 5 = strongly agree. The internal consistency reliability coefficient was α = .70. At postintervention, children also rated perceived support from their cross-age mentor or adult leader.

Self-efficacy for healthful eating and physical activity was measured by an 8-item scale; 5 items targeted eating behaviors (“I am sure I am able to … ”) defined earlier in current lifestyle behaviors and scored from 1 to 5, with 5 = strongly agree and 1 = strongly disagree. The internal consistency reliability coefficient was α = .72. Behavioral specific self-efficacy has been found to predict behavioral intention and actual behaviors (Bebestos et al., 2002; Brugg, deVet, deNooijer, & Verplanken, 2006). Three items targeted the behavior of being physically active, as defined earlier in current lifestyle behaviors.

Intention to eat healthfully and engage in physical activity was measured by an 8-item Likert-type scale (“I plan to …), five for each of the target diet behaviors of eating (a) at least one fruit daily, (b) healthy foods at lunch, (c) healthy foods at dinner, (d) healthy foods 5 or more days of the week, and (e) breakfast daily; 3 items were for the target physical activity behavior of doing something on your own time, outside of what you have to do in school that makes your heart beat faster, makes you out of breath, and makes you sweat for at least 20 min each day (National Institute of Child Health and Human Development, 2003). The term “free or light foods” was defined and used in place of the term “healthy foods.” The internal consistency reliability coefficient was α = .71.

Power Analysis and Random Assignment

Separate power calculations were performed for all outcomes. In each case, a model was created and included many factors that may influence the outcomes including time, assigned group, how time and group assignment may influence each other as well as the effects from the mentor-mentee interactions. Past data were used to determine the standard deviations, the mentor-to-mentor variation, and the student-to-student variation needed to estimate the sample size. Using these estimates, the desired effect would be detected with 40 subjects in each group at 80% power. The power calculations were performed using a Monte Carlo approach (Muthen & Muthen, 2002). To account for a 20% attrition rate, a total of 116 third and fourth grade subjects or 58 children in each group were needed.

The random assignment of subjects to the two groups controlled for cofounders such as parental influence on behavior. Using BMI z score as a stratifying variable, all children were randomly assigned to either the control or the intervention group using SPSS v17.0. After obtaining consent and assent, children were assigned an identification number. The identification numbers were entered into SPSS along with grade, gender, school, and an indicator of whether the child was classified as “high BMI z score” (greater than 0) or “low BMI score” (below 0). The sample was equally split while controlling for grade, gender, school, and BMI z score classification. The cases “selected” by SPSS were placed in the mentored group and the remaining cases were assigned to the adult-led group.

Statistical Analysis

All study variables were analyzed using measures of central tendency (e.g., means, medians), frequency distributions, and standard deviations. Current lifestyle behaviors sub-scales, self-efficacy, and intention scales were analyzed using the mean score on all answered items as the subject’s overall score. Subscale scores were calculated for current eating behaviors and current physical activity behaviors. Missing values found on the attitudes scales had the mean value for answered items on that scale substituted for missing values. The observed and imputed values were then summed to obtain a total score. Perceived support from teen mentors or adult teachers was not asked at preintervention, resulting in a 6-item measure at preintervention and a 12-item measure at postintervention. Effect sizes (ESs) were calculated for outcomes and mediators. ESs measure the difference in the measure between the teen mentoring and adult teacher groups. BMI z scores were calculated using calculated BMI values from the baseline measures.

The data collected at baseline and postintervention were analyzed using multilevel linear models. Each outcome measure (e.g., BMI, blood pressure, current lifestyle behaviors) was modeled separately. The independent variables were assigned group, attendance, school that child attends, specific mentor/adult teacher, and each child. Grade, gender, and baseline BMI z score were included as covariates to account for the stratification in the experimental design. The group and attendance were interacted; the significance tests of the coefficients associated with this interaction directly addressed Hypothesis 1. In addition, the mentor/adult random effect was interacted with attendance to account for different improvement effects between mentors. Given the interest in improvement from the baseline to postintervention, one-sided t tests on the intervention effect were conducted.

For multilevel linear models constructed to address Hypothesis 2, covariates (e.g., grade, gender, and baseline BMI z score) were included, although the baseline BMI score was not a covariate in the model for BMI. Multilevel linear models constructed for Hypothesis 3 included the three covariates, and the interaction of group. Finally, multilevel linear models constructed for Hypothesis 4 were augmented with the number of sessions attended. This variable was entered as a moderating covariate and interacted with group. The significance of the coefficients associated with these interactions. The effects of group, random school effects, random effects of mentors/adult group leaders, and necessary interactions were included in the models as was done in the models for Hypotheses 1–3.

Results

After the 8-week intervention, only the teen-mentored group showed a greater increase in physical activity behavior (p = .04) and a marginal decrease in BMI (p = .06). A medium ES, defined as an ES of at least .5 but less than .8, was found for current eating behavior (ES = .57) and improved diastolic blood pressure (ES = .56) in the teen-mentored group. A change in systolic blood pressure was not found. See Table 2 for calculated means and standard deviations for variables of interest and outcomes. These results indicate that the teen-mentored group demonstrated improved health outcomes at the conclusion of the intervention. The adult teacher group did not demonstrate any improved health outcomes at postintervention.

Table 2.

Number, Means, and Standard Deviations of Outcome Variables Between Groups

Outcomes or Mediators	Preintervention		Postintervention
	N	M(SD)	N	M(SD)
Outcomes
Current healthy eating
TM	65	3.67 (74)	67	3.88 (.81)
AL	76	3.65 (73)	77	3.84 (.73)
Current physical activity
TM	65	4.01 (.87)	67	4.21 (.89)
AL	76	4.15 (.89)	77	4.17 (.91)
Body mass index
TM	67	20.42 (4.78)	67	20.16 (4.78)
AL	80	20.16 (5.26)	77	20.49 (4.97)
Systolic blood pressure
TM	68	100.55 (8.89)	67	101.48 (9.51)
AL	81	101.22(10.30)	77	103.97 (10.91)
Diastolic blood pressure
TM	68	64.39 (8.83)	67	63.32 (7.48)
AL	81	62.39 (9.86)	77	63.54 (7.64)
Mediators
Nutritional knowledge
TM	66	11.80 (1.91)	67	11.89 (2.00)
AL	77	12.17 (1.64)	77	11.85 (2.07)
Attitudes toward eating healthfully
TM	69	41.89 (5.75)	67	41.94 (6.15)
AL	81	40.47 (7.08)	77	40.91 (6.89)
Attitudes toward physical activity
TM	69	41.61 (6.42)	67	41.96 (6.85)
AL	81	42.57 (6.40)	77	41.38 (7.26)
Perceived support§
TM	68	12.63 (1.90)	67	25.72 (4.98)
AL	79	12.53 (2.57)	77	24.33 (4.79)
Self-efficacy
TM	69	15.94 (2.77)	69	16.08 (2.47)
AL	81	15.69 (3.27)	81	16.33 (3.08)
Intention
TM	68	3.94 (0.75)	67	4.05 (0.78)
AL	79	3.98 (0.80)	77	4.03 (0.89)

Note. TM = teen mentor to individuals; AL = adult-led group.

^§Preintervention mean was for the 3-item measure; postintervention mean was for the 6-item measure.

After adjusting for grade, gender, and BMI z score, the teen-mentored group had a greater positive change in intention than the adult teacher group (p = .04) and a marginal increase in nutritional knowledge (p = .08). In the teen-mentored group, medium ES was found for perceived support (ES = .60), indicating some support for a mediating effect of this variable on behavioral and health outcomes. Attitude toward physical activity and self-efficacy had small nonsignificant effects in the opposite direction of what was hypothesized. No improvement in outcomes or mediator variables was detected in the adult teacher group.

Retention rate for all children was 92%. Nearly all children who started the 8-week intervention finished the intervention. Although not all children attended every session, children in both groups attended on average over 70% of sessions. A greater number of sessions attended was found to be associated with greater increase in nutritional knowledge (p = .04) and current physical activity (p = .02) in the teen-mentored group. Greater nutritional knowledge was associated with a greater number of sessions attended in the adult teacher group (p = .03).

Differences in mediating variables of interest were detected between the schools, regardless of group assignment. At preintervention, children at School A had significantly less nutritional knowledge compared to School B and School C (p = .002). At postintervention, compared to School B, School C showed greater improvement intention (p = .01) and perceived support (p = .05). The same trend was found when comparing School C to School A, where School C had greater improvement at postintervention in intention (p = .01) and perceived support (p = .02). No differences were found based on school between School A and School B. No differences in child outcomes between the individual mentors and adult teachers were found. In other words, individual mentors or adult teachers did not make a difference in child outcomes.

Discussion

Teen mentoring of younger children is an approach underutilized to impact the pervasive problem of childhood obesity and promoting behavioral-related lifestyle patterns especially in underserved populations such as Appalachia. To date, most outcome evaluations of mentoring have focused only on satisfaction levels and not on behavioral or health as outcomes (Hanson, 2003; Sawyer, 2001). However, Karcher, Davis, and Powell (2002) found that using teens as mentors was equally effective compared to adult mentors in promoting school connectedness in young children. Sheehan et al. (1999) found a positive effect on mentee’s classroom behavior and attitudes toward risk behaviors. Westerman (2002) found that in a rural Appalachian area, teen mentors helped mentees achieve improved academic achievement and positive connectedness to parents and family. Finally, Smith (2011) found that teen mentors helped mentees improve nutritional knowledge, attitudes toward eating healthfully, and intention to eat healthfully.

Schools are an obvious setting for the implementation of teen mentoring. To date, most mentoring activities have focused on improving academic performance or reducing risk behaviors. Less focus has been on impacting lifestyle patterns of children through mentoring. The examination of a teen-mentoring delivery method to health curricula led to improvements for children in nutritional knowledge, physical activity behaviors, and health outcomes including BMI and blood pressure. Although an improvement in BMI was not expected after the 8-week intervention, results suggest that the change may be the result of increased daily physical activity in the mentored group. Perhaps, the teens were more effective role models for physical activity or more supportive of children engaging in activities and play. Consequently teen mentoring is gaining recognition as an effective and efficacious approach to impact the lives of children.

This study builds on the previous work by Smith (2011) that showed improved intention to change dietary and activity behaviors. Smith did not examine current behaviors or health outcomes. Our results suggest that in the short-term, teen mentoring to younger children is a powerful and efficacious approach to delivering health curricula and promoting healthy lifestyle patterns, especially physical activity behaviors. Although the adult teachers in this study were school employees such as librarians and school program personnel, the children in the adult teachers’ group (comparison group) did not demonstrate the same gains in healthy lifestyle behaviors or health outcomes. Despite not sharing the experience of having a personal relationship with the children at baseline or working closely with younger children as the adult teachers did, our results indicate that teen mentoring was more effective at impacting short-term lifestyle patterns and health outcomes. This result may signify the need for programming using multiple modalities of providing education and support when striving to impact healthy lifestyle behaviors. Compared to the adult-led children, the individually mentored children perceived stronger support to change lifestyle behaviors. The teen-mentoring approach’s advantage over traditional adult-led groups is the enhanced learning and support for behavioral change that results from the perceived social support and psychological safety that is promoted by using teens as mentors (Keller & Pryce, 2010). Whereas adults possess authority by virtue of age, position, and responsibility, teens provide younger children perceived psychological safety and a social network.

Our findings also indicate that attendance was important but not for all mediators or outcomes. In both groups, children who attended the most sessions had the greatest gains in nutritional knowledge. Children in the mentored group, who attended the most sessions also had the greatest gains in physical activity. Surprisingly, a dose effect was not found for other mediators or outcomes. Understanding the dose needed to have a meaningful impact on behavior warrants further study. Finally, tailored support that considers the school environment and context should be considered when implementing any health curricula. Differences between the schools indicate that intervention’s effects differed at each school. In our study, regardless of group, children from the most rural, insular, and disadvantaged school demonstrated the most gains in intention and perceived support. Utilizing health programming that best serves each school may have a greater impact than utilizing the same curricula or programming district-wide or statewide.

Limitations

This study has several limitations. First of all, even though healthy eating and increased physical activity was promoted and encouraged, children may not have engaged in healthy eating or physical activity outside of the structured program. Second, for this program, the intervention was adapted and restructured from a 10-week program into an 8-week program, a length of time that may not result in a long-term behavioral change in physical activity or dietary patterns. Third, behavioral change was only being measured for a short period after the intervention. The true long-term impact of the intervention is not fully understood. Fourth, this study compared two different delivery methods using the same curriculum. Consequently, this study utilized a comparison group rather than a true control group. Furthermore, children were not matched to their assigned mentor or adult teacher according to gender or age, thus limiting the generalization of the findings. Finally, teens selected as teen mentors varied in their own personal health behaviors and health status. It is not known whether the teen’s health status or own health behaviors impacted the mentoring relationship with their assigned child/children.

Implications for the School Nurse

Our results have implications for the school nurse school-based health services. Based on the results from the adult teacher group, the method of curricular delivery is important. Modalities of delivery that utilize individualized support rather than group support were more powerful. With training, guidance, and support, teenagers can serve as effective health coaches and role models in delivering essential health curriculum to younger children. Thus, teen mentors and effective mentoring can positively impact health behaviors and health outcomes of younger mentees.

By relying on local community residents to deliver important health curricula, the teen-mentoring approach provides insular, remote, and underserved communities with a promising method to impact the health of children. Given that teen mentoring has been shown to be effective at impacting academic outcomes of other disadvantaged children such as ethnic minority children and children of families who lack sufficient resources, the mentoring approach may be adapted by schools striving to meet the health needs of students during economically and academically challenging times. Finally, the mentoring approach was an effective approach in the delivery of health curriculum outside of the normal school day. For schools striving to meet academic standards and increasing demands on time, supported teen mentoring may be a feasible approach to continue meeting the health and lifestyle needs of children while increasing support for behavioral change but not detracting from the normal school hours.

Biographies

Laureen H. Smith, PhD, RN is an associate professor at the Ohio State University.

Christopher Holloman, PhD, is an associate professor at the Ohio State University.