CHILDHOOD OBESITY

February 2014 j Volume 10, Number 1

ORIGINAL ARTICLES
ª Mary Ann Liebert, Inc.
DOI: 10.1089/chi.2013.0049

Long-Term Obesity Prevention and the

Motivating Adolescents with Technology
to CHOOSE Health Program
Suzanne Lazorick, MD, MPH,1,2 Yancey Crawford, MPH,1 Anthony Gilbird, MPH,3
Xiangming Fang, PhD,4 Veronica Burr, MPH,5
Valeria Moore, MPH,6 and George T. Hardison, Jr, MA1

Abstract
Background: The Motivating Adolescents with Technology to CHOOSE Health (MATCH) intervention integrates lifestyle
behavior change curriculum within academic subjects taught in seventh grade. This study assesses obesity prevention in participants
into high school.
Methods: The study compares four- to five-year longitudinal data from a single-site cohort (N = 106, 54% retained from 195
participants at baseline; 82% of those still at the school) pre- and postintervention in a rural middle school with high obesity rates
with data from the 2006 Child Survey and 2010 Child and Young Adult Surveys from the National Longitudinal Survey of Youth
1979 (N = 600), which serves as a nationally representative comparison group. Outcome measures include pre- and postchanges in
weight category, BMI, BMI z-score, BMI percentile for age and gender, and rates of change per month in BMI measures.
Results: At follow-up, change in percent overweight was significantly different between groups, with the MATCH group de-
creasing (20–12%) and the comparison group increasing (17–19%). Overall, the MATCH group had significantly higher decrease
rates in BMI z-scores ( p = 0.002) and BMI percentile ( p = 0.01) than the comparison group. Of all adolescents at healthy weight at
baseline, 2% from MATCH became overweight after five years, whereas 13% of the comparison group increased to overweight or
obese ( p = 0.02) after four years.
Conclusions: Despite a small sample size and a high-risk setting, at long-term follow-up, a greater proportion of MATCH
participants than in the comparison group decreased from overweight to healthy weight or remained at healthy weight. The MATCH
results suggest that some proportion of high-risk adolescents can have their growth trajectory follow a healthier path than expected.

Introduction nutrition and increased physical activity,3,4 policy change

can be slow and difficult. The need remains for effective,
hildhood obesity is a public health epidemic. In the scalable, intensive interventions to influence lifestyle be-

C United States, recent data reveal that 16.9% of

children ages 2–19 years are obese, with obesity
increasing nationally in all age groups under 19 since
haviors and prevent obesity. Schools are key settings for
public health strategies based on the consistent, intensive
contact teachers and staff have with children5 and because
1976.1 Some recent regional studies suggest that the of opportunities to engage children in healthy eating and
prevalence of childhood obesity may be stabilizing2; physical activity and teach, model, and reinforce wellness
however, prevalence remains unacceptably high. Although messages.6 Young adolescents are developing decision-
there is increasing momentum for policy and environ- making skills and building lifelong habits, making middle
mental changes to reduce obesity by supporting improved school an opportune setting for school-based intervention.6

1
Department of Pediatrics, East Carolina University Brody School of Medicine, Greenville, NC.
2
Department of Public Health, East Carolina University Brody School of Medicine, Greenville, NC.
3
Office for Human Research Integrity, East Carolina University Brody School of Medicine, Greenville, NC.
4
Department of Biostatistics, East Carolina University College of Allied Health Sciences, Greenville, NC.
5
Veritas Collaborative, Durham, NC.
6
United States Food and Drug Administration, Office of Regulatory Affairs, St. Louis, MO.

25
26 LAZORICK ET AL.

Large school interventions in the 1990s, such as the sample with longitudinal BMI measures recorded at
Child and Adolescent Trial for Cardiovascular Health comparable intervals.
(CATCH) and Planet Health, were designed to combat
cardiovascular risk and/or obesity and have shown promise Methods
in influencing knowledge, habits, and, to some degree,
weight.7,8 Recently, the HEALTHY study, a large, rigor- A single-site cohort pre- and postintervention study with
ous, school-based intervention, has also shown promising longitudinal follow-up (MATCH) was compared to a na-
results for decreasing obesity-related risk factors for met- tionally representative sample of similar-age children with
abolic disease, albeit without significant increases in height and weight recorded at near-concurrent intervals. The
prevalence of healthy weight, compared to control.9 The MATCH intervention has been described previously,12 and
CATCH trial demonstrated successful sustained changes in key components of the model are shown in Table 1.
select health behaviors, but not physiologic measures.10 MATCH integrates a health, nutrition, physical activity, and
However, to date, interventions based in the middle-school technology curriculum with the North Carolina Standard
years have not shown sustained intervention effects be- Course of Study for seventh-grade students to achieve
yond young adolescence, and intervention effects achieved health-related outcomes while simultaneously meeting state
during the school year are often lost over the summer.11 and national educational objectives. It is designed to be
Motivating Adolescents with Technology to CHOOSE scalable and requires no additional school staff, minimal
Health (MATCH) is a teacher-developed, school-based training, and low-cost additional materials (e.g., a work-
childhood obesity intervention for seventh graders that book, pedometers, and small incentive items). It is aligned
began in 2007 in a rural, high-minority, economically with the CDC school model, the Coordinated School Health
challenged middle school in eastern North Carolina. Re- Program,13 and follows published recommendations for
sults immediately after the intervention and two years achieving a healthy weight.14 MATCH included 55 contact
later were promising (88% and 99% participation rates hours provided over a period of 14–16 weeks and was
each year; within the subset, among all overweight par- specifically designed to reach rural, underprivileged, and
ticipants, BMI z-score changes were - 0.08 and - 0.04 minority youth.12
each year, respectively).12 The aim of this study is to
examine long-term outcomes for the first two cohorts of Intervention Group
adolescents participating in MATCH. Specifically, this In the 2006–2007 and 2007–2008 school years, all stu-
study explores results of MATCH participation for obe- dents enrolled in regular seventh-grade classes at one mid-
sity prevention by examining the changes in proportion of dle school in eastern North Carolina completed MATCH
adolescents in each weight category at baseline and after within routine school activities, and those assenting and
four and five years and the rates of change in BMI mea- with parents consenting were included in the research study
sures over time, compared to a nationally representative (N = 195, 93% average participation rate). The original

Table 1. Components of the Motivating Adolescents with Technology to CHOOSE HealthTM

(MATCH) Intervention Educational Model
Key element of the MATCH educational model Example intervention components
Conceptual knowledge (learn ‘‘why’’) 14-week interdisciplinary wellness lessons and activities embedded
within standard curricula
Health skills (learn ‘‘how’’)  Daily physical activity, pedometers
 Self-tracking of physical activity
 Food intake record and analysis
 Energy balance activities
Individualized tasks (apply the skill)  Calculate BMI, determine weight category
 Fitness testing
 Self-evaluation of health behaviors
Motivational strategies (reinforce the skill)  Individual goal setting with action plans
 Peer accountability contracts
 Recognition bulletin board
 Incentive items for achieving goals
CHILDHOOD OBESITY February 2014 27

study and intervention have been described previously.12 mographics (race, gender, and year of birth), age in months
The school is located in a rural North Carolina county where (child survey), age in years (young adult survey), date of
24% of residents live in poverty and 43% are African survey, height (inches), weight (pounds), and method of
American.15,16 Nearly all of the students attending this height and weight assessment (measurement, mother report,
middle school go on to attend one high school. and self report) were queried for children between 120 (10
For this study, in January 2012, all 11th- and 12th-grade years) and 179 (14.9 years) months of age at the time of their
students (3.5 and 4.5 years post-MATCH, respectively) at interview in 2006. The 680 children with birth years of
the high school were invited by letter to have their height 1993–1995 and having both height and weight values re-
and weight measured at school. An ‘‘opt-out’’ consent/ corded in 2006 were eligible for inclusion; 600 of these also
assent procedure was used, such that all students from the had height and weight values for 2010 and were included for
original study who were remeasured and did not return a analysis. This group is not intended to serve as a substitute
signed ‘‘opt-out’’ form were included for analysis. To in- for a ‘‘control’’ condition; rather, it represents changes in
crease the likelihood of participation, an incentive, con- height and weight that reasonably would be expected in a
sisting of the opportunity to attend a MATCH celebration diverse group of children without the MATCH intervention
with the availability of an individual-portion ice-cream over time between ages 10 and 15 years.
snack and drawings for backpacks, water bottles, and
drawstring bags, was offered. Of the original 195 MATCH Statistical Analysis
participants who were expected to be in the 11th or 12th Because the study focus is changes at follow-up, only
grade, 66 were no longer at the school. Of the 129 eligible, adolescents with both baseline and follow-up measure-
nine opted out and 14 were either absent or declined re- ments were included. Baseline characteristics of those in-
measure. The final intervention group includes the 106 cluded and excluded were compared for each group using
students remaining from MATCH (54% of the original two-sample t-tests for continuous variables (age, BMI,
cohorts, 82% of those available). BMI percentile, and BMI z-score) and chi-square tests (or
Gender, ethnicity (provided by parent upon school reg- Fisher’s exact test, as appropriate) for categorical variables
istration), and birth date were recorded from school files. (gender and race). Similar statistical methods were also
Age was calculated from dates of birth and measurement. employed for comparing baseline characteristics between
Height and weight measures (once at each time point with the MATCH and comparison groups.
shoes off, wearing the standard school uniform; using cali- Two-way tables were used to describe the weight-
brated scale) were done privately following a defined pro- category distributions at baseline and follow-up for each
tocol. A Schorr stadiometer (Schorr Productions, Olney, group. Because of the nature of repeated-measure data, a
MD) was used for all height measures, except for in spring generalized linear model with the GEE (generalized esti-
2007. A school nurse measured students in the intervention mating equations)20 method was used to study the per-
year. Follow-up measures were collected by a trained re- centage change of each weight category from baseline to
search team. BMI was calculated from height and weight follow-up and the difference between the two groups.
and BMI z-score and BMI percentile for age and gender To account for differences in the length of follow-up time
were determined from the standardized CDC charts. Weight between the MATCH and comparison groups, outcome
category was assigned based on current CDC definitions.17 variables include changes per month in BMI, BMI percen-
tile, and BMI z-score. These change rates per month were
Comparison Group compared between the two groups using both two-sample t-
The National Longitudinal Survey of Youth 1979 cohort tests and multiple regression models to control for other
(NLSY79) is a multi-purpose panel survey sponsored by effects. The same comparisons were also done within each
the Bureau of Labor Statistics, US Department of Labor.18 weight category. To further investigate how BMI changes
All data from NLSY are deidentified and available for among adolescents who were at the upper end of the healthy
research. The original cohort included a nationally repre- weight range at baseline, spaghetti plots were created to
sentative sample of 12,686 men and women, oversampled depict the BMI percentile change for each adolescent
for African American and Hispanic groups, who were 14– with 70th to < 85th BMI percentile at baseline. Repeated-
21 years of age on December 31, 1978. Starting in 1986, measures analysis of variance was used to compare the
the children (NLSY79 Child Survey) of female respon- mean trend between the MATCH and comparison groups.
dents have been interviewed and assessed for height and This study was approved by the university medical
weight every two years. At baseline, these were either center institutional review board (#07-0741) at the Brody
measured by a nurse or the mother (53.3%) or reported by School of Medicine, East Carolina University (Greenville,
the mother (46.7%). At follow-up (age ‡ 14 years), height NC).
and weight were by adolescent self-report (NLSY79
Young Adult Survey).
Data from the 2006 NLSY79 Child Survey and the 2010
Results
NLSY79 Child and Young Adult Surveys were obtained Table 2 shows the characteristics of the MATCH and
from the NLS Investigator website.19 System ID, basic de- comparison groups at baseline and demonstrates
28 LAZORICK ET AL.

cents being overweight or obese and none underweight; in

Table 2. Baseline Participant contrast, the comparison group had just over one third
Characteristics, MATCH Intervention overweight or obese and 5% underweight. In long-term
and NLSY Comparison Groups follow-up, the weight category measures tended to im-
Intervention group Comparison group prove more in the MATCH group than in the comparison,
MATCH NLSY79 young although differences between the groups only reached
adults
N5106 N5600
statistical significance in the overweight category. The
mean (SD) mean (SD) percent healthy weight increased significantly in both
Age (months) 153.5 (5.78) 145.4 (10.09) groups, with MATCH increasing from 49% to 63% (in-
crease of 14%), compared to 59–67% (increase of 8%) in
BMI 23.78 (6.08) 21.11 (5.42) the comparison group. Whereas the percent overweight in
BMI percentile 76.71 (24.41) 65.04 (30.96) the MATCH group decreased from 20% to 12%, in the
BMI z-score 1.02 (0.99) 0.52 (1.33) comparison group it increased from 17% to 19%. The
percent obese decreased in both groups by approximately
N (%) N (%)
6% (MATCH decreased from 31% to 25%; comparison
Gender decreased from 20% to 14%). The generalized linear
Female 60 (57) 282 (47) model with the GEE method for each weight category
Male 46 (43) 318 (53) showed that the increase in percent healthy weight and the
decrease in percent obese were statistically significant for
Race both groups ( p < 0.001), but no significant difference in
Nonblack/non- 39 (37) 317 (53) these changes was found between the intervention and
Hispanica comparison groups. Within each group, the change in
Black 64 (60) 167 (28) percent overweight was not statistically significant, but the
change was found to be significantly different between the
Hispanic 1 (1) 116 (19)
two groups ( p = 0.03).
Asian 2 (2) — Changes per month in BMI, BMI percentile, and BMI z-
a
NLSY categorizes race as black, Hispanic, or nonblack/non- score were compared between the MATCH and compari-
Hispanic; participants in the MATCH group were categorized as son groups (Table 5). Overall, the MATCH group had
white, black, Hispanic, or Asian. significantly higher decrease rates in BMI z-score
MATCH, Motivating Adolescents with Technology to CHOOSE ( p < 0.002) and BMI percentile ( p = 0.01). Within the
Health; NLSY, National Longitudinal Survey of Youth; SD, healthy weight category, the MATCH group was found to
standard deviation. have a significantly lower increase rate in BMI ( p < 0.001)
and significantly higher decrease rates in BMI z-score
( p = 0.01) and BMI percentile ( p = 0.02). For the obese
category, the MATCH group had a higher increase rate in
BMI ( p = 0.02) and lower decrease rates in BMI percentile
considerable differences. The MATCH cohort was older ( p = 0.04) and BMI z-score ( p = 0.02). Multiple regression
by 8.1 months, had higher mean BMI measures, greater models were also attempted to control for effects of gen-
percent female (57% vs. 47%), higher proportion black der, ethnicity, and age in the above comparisons, but none
(60% vs. 28%), and almost no Hispanic participants of these were found to be statistically significant. There-
( < 1%), compared to the comparison group (19%). fore, only results from two-sample t-tests are reported.
For each group, baseline characteristics were compared Figure 1 depicts the weight trajectory over time for the
for those excluded versus included for final analysis for subset of individuals from the upper end of the healthy
both the MATCH (n = 89 lost, 106 retained) and compar- weight category from the MATCH and comparison groups
ison (n = 80 excluded; n = 600 included) groups (Table 3). who, at baseline, measured between the 70th and < 85th
In the MATCH group, those retained were significantly BMI percentile. The mean trajectory is shown for each
younger at baseline, with a mean age 4.9 months younger group with a bold line and reveals a steeper decline in the
than those lost to follow-up ( p < 0.001). There were no MATCH group ( p = 0.01); this significance level was
significant differences in age in the comparison group, as maintained even after exclusion of the student in the
well as in either group for BMI measures, gender, and race MATCH group who had the largest decrease in BMI per-
or ethnicity. centile (a drop from 84th to 13th percentile). In this healthy
Table 4 shows the proportion of adolescents in each weight subset, no adolescents (0 of 18) from MATCH in-
weight category at both baseline and follow-up for the creased to overweight, whereas in the comparison group,
MATCH and comparison groups. Consistent with the high- 20% (22 of 108) increased to overweight or obese. This
risk, rural, low-resource community in which the middle difference is statistically significant ( p = 0.04). Also, of all
school is located, at baseline, the MATCH group had adolescents at healthy weight at baseline, 2% (1 of 52)
higher prevalence of obesity, with over half of all adoles- from MATCH changed to overweight after five years,
CHILDHOOD OBESITY February 2014 29

Table 3. Baseline Characteristics of Participants Lost to Follow-Up/Excluded

and Retained/Included in MATCH and NLSY Groups
Lost to follow-up/excluded Retained/included Test for difference
mean (SD) mean (SD) t-test p value
MATCH N = 89 N = 106
Age (months) 158.4 (9.99) 153.5 (5.78) < 0.001
BMI 25.30 (6.89) 23.78 (6.08) 0.10
BMI percentile 79.10 (22.95) 76.71 (24.41) 0.48
BMI z-score 1.17 (0.96) 1.02 (0.99) 0.29
NLSY N = 80 N = 600
Age (months) 143.5 (11.15) 145.4 (10.09) 0.15
BMI 21.30 (6.00) 21.11 (5.42) 0.79
BMI percentile 64.84 (31.14) 65.04 (30.96) 0.95
BMI z-score 0.53 (1.34) 0.52 (1.33) 0.94
N (%) N (%) Chi-square test p value
MATCH
Gender
Female 50 (56) 60 (57) 0.95
Male 39 (44) 46 (43)
Race
Asian 0 2 (2)
Black 56 (63) 64 (60)
Hispanic 1 (1) 1 (1) 0.51
White 30 (34) 39 (37)
Native American 1 (1) 0
Other 1 (1) 0
NLSY
Gender
Female 44 (55) 282 (47) 0.19
Male 36 (45) 318 (53)
Race
Nonblack/non-Hispanica 48 (60) 317 (52)
Black 15 (19) 167 (28) 0.22
Hispanic 17 (21) 116 (19)
a
NLSY categorizes race as black, Hispanic, or nonblack/non-Hispanic; participants in the MATCH group were categorized as white, black,
Hispanic, or Asian.
MATCH, Motivating Adolescents with Technology to CHOOSE Health; NLSY, National Longitudinal Survey of Youth;
SD, standard deviation.

whereas 13% (45 of 353) in the comparison group became reaching statistical significance, these results suggest that
overweight or obese over four years ( p = 0.02). adolescents participating in MATCH sustained healthier
BMI trajectories over time than may have been expected,
given their high-risk environment, and nearly all who
Discussion were healthy weight at the start did not progress to over-
Despite small sample size, low retention rate in the weight. The comparison group represented expected
MATCH group, and demonstrated differences not always growth in height and weight between middle-school– and
30 LAZORICK ET AL.

Table 4. Percent in Each Weight Category at Baseline and Follow-Up, MATCH and NLSY
Comparison Group
Test of Test of
difference baseline difference between
Baseline Follow-up to follow-up groups
Weight statusa N% N% p value p value
MATCH Healthy weight 52 (49) 67 (63) < 0.001 0.33
Overweight 21 (20) 13 (12) 0.38 0.03
Obese 33 (31) 26 (25) < 0.001 0.29
Underweight 0 0 — —
Comparison Healthy weight 353 (59) 400 (67) < 0.001
Overweight 100 (17) 111 (19) 0.38
Obese 117 (20) 81 (14) < 0.001
Underweight 30 (5) 8 (1) —
a
Weight category determined by CDC definitions based on BMI percentile for age and gender: underweight, < 5th percentile;
healthy weight, 5th to < 85th percentile; overweight, 85th to < 95th percentile; obese, ‡ 95th percentile.
MATCH, Motivating Adolescents with Technology to CHOOSE Health; NLSY, National Longitudinal Survey of Youth.

Table 5. Changes per Month in BMI Measures by Baseline Weight Category

in MATCH vs. NLSY Comparison Groups
MATCH change NLSY change
Baseline weight per month per month Test of difference
category mean (SD) mean (SD) p value
All N 106 600
BMI 0.047 (0.053) 0.053 (0.078) 0.31
BMI z-score - 0.004 (0.011) 0.000 (0.023) 0.002
BMI percentile - 0.113 (0.328) - 0.018 (0.529) 0.01
Healthy weight n 52 353
BMI 0.042 (0.038) 0.062 (0.054) < 0.001
BMI z-score - 0.003 (0.013) 0.002 (0.017) 0.01
BMI percentile - 0.121 (0.435) 0.044 (0.568) 0.02
Overweight n 21 100
BMI 0.040 (0.047) 0.042 (0.086) 0.83
BMI z-score - 0.006 (0.009) - 0.008 (0.016) 0.45
BMI percentile - 0.167 (0.206) - 0.262 (0.415) 0.13
Obese n 33 117
BMI 0.058 (0.073) 0.019 (0.118) 0.02
BMI z-score - 0.004 (0.008) - 0.009 (0.015) 0.02
BMI percentile - 0.064 (0.141) - 0.142 (0.299) 0.04
MATCH, Motivating Adolescents with Technology to CHOOSE Health; NLSY, National Longitudinal Survey of Youth;
SD, standard deviation.
CHILDHOOD OBESITY February 2014 31

Figure 1. BMI percentile change over time in Motivating Adolescents with Technology to CHOOSE Health (MATCH) and National
Longitudinal Survey of Youth (NLSY) comparison groups for adolescents at 70th to < 85th BMI percentile at baseline. Bold line shows
mean change over time; slope in the MATCH group shows more rapid decline ( p = 0.01).

late-high-school–age adolescents across the nation, prevention policy and environmental change in North
whereas the MATCH group was from an economically Carolina (NC), and substantive efforts have focused on
challenged county with a high minority population at very nutrition and physical activity (PA) in elementary-
high risk of adult obesity. The question could be raised school–age children. In addition, state school board pol-
asking whether the students in MATCH were especially icy requires 30 minutes per day of PA for middle school
motivated because of some other factor or came from an students, although it is not uniformly provided. MATCH
unusually supportive environment for behavior change. participants did participate in PA of 25 minutes per day in
However, this school district is not known to have highly grades 6–8, but not after. It seems that effects of these
involved Parent-Teacher Associations, energized faculty, state-wide efforts should have been most evident at
additional physical activity or nutrition curricula, or other baseline and are unlikely to have had a sufficient effect on
county-wide public health efforts that would have sup- Martin County students to produce the promising results
ported the students to reach healthier BMI trajectories. described. Similar increases in the proportion of older
One possible factor that may have contributed to the teens at healthy weight have not been described in other
noted results, but also represents a potential barrier to areas of the state.22,23
replication, is the intensity of the curriculum (55 contact This study has several limitations that have been de-
hours, of which approximately 20 were nutrition content, scribed previously.12 The results are from one school with
which is 5-fold greater than the 4.2 hours of nutrition no control group, and the demographics of the school are
education that is typically provided in middle schools, not representative of urban, high-income, Hispanic, or
according to national survey data from the CDC).21 It is Asian populations, so results may not be generalizable. No
also possible that the promising results are because of behavioral measures were completed pre- and post-
other statewide activities designed to reduce childhood MATCH, so the underlying possible mechanisms for the
obesity. Eat Smart Move More North Carolina, the state’s observed differences in BMI trajectory, such as specific
obesity coalition, seeks to establish statewide obesity changes in nutrition or physical activity behaviors, cannot
32 LAZORICK ET AL.

be investigated. Given the limited design, findings should and prevented future obesity. Although, because of study
be interpreted with caution. limitations, the result cannot be attributed to MATCH
Additional limitations are specific to this follow-up participation, the fact that over 98% of participants
study. The comparison group used to represent expected who were healthy weight at baseline remained so four or
growth in height and weight in adolescents over a similar five years later suggests they achieved energy balance
time period was different in characteristics from the over an extended period, and the results were not af-
MATCH group. Because the MATCH group was from fected by either gender or race/ethnicity. Several previ-
such a high-risk population, one would expect results to be ous school-based studies, if improvements in BMI were
worse in MATCH than in the comparison group; however, achieved, found results to be inconsistent across demo-
no differences associated with gender or race/ethnicity graphic groups.7,27,28
were found. To locate additional estimates of how mean
BMI measures may be changing with age between 2006 Conclusion
and 2012 in other groups of young adolescents, available
recent sources of cross-sectional BMI measures were ex- MATCH was implemented in a school with very low
amined.2,24 When looking at mean measures in middle resources and with students perceived as being at the
school or seventh-grade students and comparing to high highest risk for obesity. The curriculum and activities are
school or 11th- and 12th-grade students, in general, the theoretically driven, having been intentionally designed to
highest increases reported in the proportion of healthy apply social cognitive theory of self-regulation,29 and fit
weight students from schools ranged from 5% to 8% and within the standard course of study; thus, the curriculum
thus, overall, were comparable to that found in the NLSY may be adapted to other settings without substantial ad-
comparison group. ditional personnel or materials. After expansion to three
A final limitation is the relatively low retention rate of and then six schools, MATCH is being implemented in 13
54% in the MATCH group, although most of those ado- schools in North Carolina and four in South Carolina in the
lescents not remeasured had moved out of the area or the 2013–2014 school year. Future research is needed to rig-
school, as opposed to dropping out of the study. No sta- orously test this model to prove effectiveness with con-
tistically significant differences in baseline characteristics trolled study, assess for accompanying behavior changes,
in those retained versus lost were found, even in BMI investigate underlying mechanisms, and, if appropriate,
measures. Although no systematic differences that would further develop dissemination strategies to schools across
affect results in those lost to follow-up were identified, it different regions.
may have been that those lost were somewhat more over- Strategies to address obesity will need to be broad based
weight at baseline. Given the small sample size, if those across community sectors, but also adaptable. Using a school-
lost to follow-up had worse weight trajectories than those based model such as MATCH holds promise because
retained, then the results in the MATCH group may be once core educational elements and activities are identi-
overestimating intervention effects. fied, if adopted, they have the potential to reach substantial
This study also has several strengths. Preceding fairly numbers of youth and prevent development of obesity.
recent, routine school-based measures of children, there
have been few available data sets or obesity studies with Acknowledgments
longitudinal, measured height and weight measures. The
sources most commonly cited for national prevalence es- This work was supported by funding from the Blue
timates are from self-reported data,25 and a known limi- Cross and Blue Shield of North Carolina Foundation and,
tation of the CDC growth charts that define weight in part, by the Robert Wood Johnson Foundation Physician
category is that they represent cross-sectional data.26 Al- Faculty Scholar Award Program. An abstract of the
though the comparison group anthropometric data for this MATCH results included in this article was presented as a
study were obtained from a combination of measures and poster at The Obesity Society 30th Annual Scientific
self-report, a strength of the MATCH methods is that it Meeting, San Antonio, Texas, September 20–24, 2012.
includes measured longitudinal data at up to five years The authors acknowledge the several Martin County
postintervention; to the knowledge of the authors, this is businesses that contributed a combined total of $6,500 of
the longest reported longitudinal follow-up of a school- funding and supplies for MATCH, as well as backpacks
based intervention. In addition, the MATCH group repre- that served as an incentive for participation in the follow-
sents a very high-risk population and results from an up measures. The authors also thank the Martin County
innovative, feasible strategy for obesity prevention. school system personnel who supported this work, partic-
Even considered with caution, the potential implica- ularly administrative leadership during 2006–2008, in-
tions of these findings are noteworthy. These results cluding Superintendent Dr. Thomas Daly. Finally, the
suggest that the weight trajectories of young adolescents authors recognize Principal Clay Wagner, who supported
living in a high-risk environment for obesity can be MATCH at the middle school in 2006–2008, and then
modified. Participation in MATCH for some subset of greatly facilitated the follow-up study as the high school
adolescents may have led to sustained lifestyle changes principal in 2011–2012.
CHILDHOOD OBESITY February 2014 33

Author Disclosure Statement CountyPrint.php?county = NorthCarolina Last accessed September

25, 2012.
No competing financial interests exist. 16. North Carolina Rural Economic Development Center. County
profile for rural county. 2011. Available at www.ncruralcenter.org/
databank/CountyPrint.php?county=Rural Last accessed September
25, 2012.
References 17. Centers for Disease Control and Prevention. Basics about child-
hood obesity. 2012. Available at www.cdc.gov/obesity/childhood/
1. Fryar CD, Carroll MD, Ogden CL. Prevalence of obesity among basics.html Last accessed September 25, 2012.
children and adolescents: United States, trends 1963–1965 through
18. Bureau of Labor Statistics, US Department of Labor, and National
2009–2010. 2012. Available at www.cdc.gov/nchs/data/hestat/
Institute for Child Health and Human Development. Children of
obesity_child_09_10/obesity_child_09_10.pdf Last accessed Oc-
the NLSY79, 1979–2010. The Center for Human Resource Re-
tober 21, 2013.
sarch. The Ohio State University: Columbus, OH, 2012.
2. Robbins JM, Mallya G, Polansky M, et al. Prevalence, disparities,
19. National Longitudinal Surveys. NLS information. 2011. Available
and trends in obesity and severe obesity among students in the
at www.nlsinfo.org Last accessed September 25, 2012.
Philadelphia, Pennsylvania, school district, 2006–2010. Prev
Chronic Dis 2012;9:E145. 20. Liang KY, Zeger SL. Longitudinal data-analysis using generalized
linear-models. Biometrika 1986;73:13–22.
3. Frieden TR, Dietz W, Collins J. Reducing childhood obesity
through policy change: Acting now to prevent obesity. Health Aff 21. Kann L, Telljohann SK, Wooley SF. Health education: Results
(Millwood) 2010;29:357–363. from the School Health Policies and Programs Study 2006. J Sch
Health 2007;77:408–434.
4. Friedman RR, Schwartz MB. Public policy to prevent childhood
obesity, and the role of pediatric endocrinologists. J Pediatr 22. North Carolina Child Health Assessment and Monitoring Program
Endocrinol Metab 2008;21:717–725. (CHAMP) survey data 2007–2010 regional results. Raleigh, NC:
State Center for Health Statistics, 2007–2008, 2008–2009, 2009–
5. Peterson KE, Fox MK. Addressing the epidemic of childhood
2010.
obesity through school-based interventions: What has been done
and where do we go from here? J Law Med Ethics 2007;35: 23. North Carolina Child Health Assessment and Monitoring Program
113–130. (CHAMP) survey data 2008–2011 statewide results. Raleigh, NC:
State Center for Health Statistics, 2008, 2009, 2010, 2011.
6. Chassin L, Presson C, Sherman S, et al. Adolescent Health Issues.
2nd ed. Guilford Press: New York, 1995. 24. Arkansas Center for Health Improvement. Year Eight Assessment
of Childhood and Adolescent Obesity in Arkansas (Fall 2010–
7. Gortmaker SL, Peterson K, Wiecha J, et al. Reducing obesity via a
Spring 2011). AR ACHI: Little Rock, AR, February 2012.
school-based interdisciplinary intervention among youth: Planet
Health. Arch Pediatr Adolesc Med 1999;153:409–418. 25. Centers for Disease Control and Prevention. Behavioral Risk
Factor Surveillance System survey data (obesity trends). Atlanta,
8. Luepker RV, Perry CL, McKinlay SM, et al. Outcomes of a field
GA: US Department of Health and Human Services, Centers for
trial to improve children’s dietary patterns and physical activity.
Disease Control and Prevention, 2011.
The Child and Adolescent Trial for Cardiovascular Health.
CATCH collaborative group. JAMA 1996;275:768–776. 26. Grummer-Strawn LM, Reinold C, Krebs N. Use of World Health
Organization and CDC growth charts for children aged 0–59
9. Kaufman F, Hirst K, Buse J, et al. Effect of secular trends on a
months in the United States. Report no. RR-9. Atlanta, GA:
primary prevention trial: The HEALTHY study experience. Child
Centers for Disease Control and Prevention, 2010.
Obes 2011;7:291–297.
27. Franks A, Kelder SH, Dino GA, et al. School-based programs:
10. Nader PR, Stone EJ, Lytle LA, et al. Three-year maintenance of
Lessons learned from CATCH, Planet Health, and Not-On-To-
improved diet and physical activity: The CATCH cohort. Child
bacco. Prev Chronic Dis 2007;4:A33.
and Adolescent Trial for Cardiovascular Health. Arch Pediatr
Adolesc Med 1999;153:695–704. 28. Wiecha JL, El Ayadi AM, Fuemmeler BF, et al. Diffusion of an
integrated health education program in an urban school system:
11. Carrel AL, Clark RR, Peterson S, et al. School-based fitness
Planet Health. J Pediatr Psychol 2004;29:467–474.
changes are lost during the summer vacation. Arch Pediatr Adolesc
Med 2007;161:561–564. 29. Bandura A. Social Foundations of Thought and Action: A Social
Cognitive Theory. Prentice-Hall: Englewood Cliffs, NJ, 1986.
12. Lazorick S, Hardison GT, Esserman DE, et al. Sustained body
mass index changes one and two years post MATCH: A school-
based wellness intervention in adolescents. Child Obes 2011;7:
372–378. Address correspondence to:
13. Centers for Disease Control and Prevention. How schools can im- Suzanne Lazorick, MD, MPH
plement coordinated school health. 2011. Available at www.cdc Associate Professor of Pediatrics and of Public Health
.gov/healthyyouth/cshp/schools.htm Last accessed January 23, 2013. East Carolina University Brody School of Medicine
14. Barlow SE. Expert committee recommendations regarding the pre- 600 Moye Boulevard
vention, assessment, and treatment of child and adolescent over-
weight and obesity: Summary report. Pediatrics 2007;120(Suppl. 4): 174 Life Sciences Building
S164–S192. Greenville, NC 27834
15. North Carolina Rural Economic Development Center. Profile for
North Carolina. 2011. Available at www.ncruralcenter.org/databank/ E-mail: [email protected]