Available via license: CC BY-NC-ND
Content may be subject to copyright.
RESEARCHARTICLE
Effect of Delaying High School Start Time on
Teen Physical Activity, Screen Use, and
Sports and Extracurricular Activity
Participation: Results From START
AARON T. BERGER, PhD, MPHaDARIN J. ERICKSON, PhDbKAYLA T. JOHNSON, PhDcEMMA BILLMYER,MS
dKYLA WAHLSTROM, PhDe
MELISSA N. LASKA, PhD, RDfRACHEL WIDOME, PhD, MHSg
ABSTRACT
BACKGROUND: We aimed to characterize relationships between delayed high school start time policy, which is known to
lengthen school night sleep duration, and patterns in activity outcomes: physical activity, non-school electronic screen time
(non-schoolwork), and sports and extracurricular activity among adolescents.
METHODS: We used data from the START study, a multi-site evaluation of a natural experiment, assessing the effects of a
school start time policy change in high schools in the Minneapolis, Minnesota metropolitan area. The study follows students in
2 schools that shifted to a later start time (8:20 or 8:50 AM) after baseline year and 3 schools that maintained a consistent, early
start time (7:30 AM) over the 3-year study period. Activity was measured by participant self-report on an in-school survey. The
analysis used a difference-in-differences estimator, in which the changes in each outcome observed in the comparison schools
estimate the changes in each outcome that would have been observed in the late-start adopting schools had they not delayed
their start times after baseline.
RESULTS: Over 2 years of follow-up, no changes emerged to suggest that later school start times either interfered with, or
promoted, any activity-related outcome that was measured.
IMPLICATIONS: Communities interested in promoting sleep by delaying start times may do so knowing that there are unlikely
to be adverse effects on adolescent physical activity, electronic screen time, or organized sports and activity participation.
CONCLUSIONS: A shift to later school start times does not appear to enhance or detract from the healthfulness of students’
activity level.
Keywords: school start time; sleep; physical activity; sedentary behavior.
Citation: Berger AT, Erickson DJ, Johnson KT, et al. Effect of delaying high school start time on teen physical activity, screen use,
and sports and extracurricular activity participation: results from START. J Sch Health. 2025; 95: 70-77. DOI: https://doi.org/10
.1111/josh.13506.
Received on March 16, 2024
Accepted on September 10, 2024
The US Department of Health and Human Services
recommends children and adolescents engage in
at least 60 minutes of physical activity daily and
aim to replace sedentary time with activity.1Yet
only roughly one quarter of high school students
overall report meeting the 60-minute daily physical
activity guideline,2and demographic groups that face
social and economic disadvantage are even less likely
to meet the recommendation.3Moreover, as youth
age through the adolescent years, physical activity
aResearch Assistant (berge314@umn.edu), Division of Epidemiology and Community Health, University of Minnesota School of Public Health, Minneapolis, MN.
bProfessor, (erick232@umn.edu), Division of Epidemiology and Community Health, University of Minnesota School of Public Health, Minneapolis, MN.
cPostdoctoral Fellow, (joh13524@umn.edu), Division of Epidemiology and Community Health, University of Minnesota School of Public Health, Minneapolis, MN.
time decreases. Further, there has been a secular
trend of adolescent physical activity decline4and
sedentary behavior time increase5in recent decades.
While a variety of individual, community, and societal
issues influence youths’ 24-hour movement behavior,
a potentially modifiable factor that has received
increasing attention in recent years is sleep.
In the United States, on school nights, the vast
majority of adolescents do not meet the 8-hour
minimum sleep duration recommendation set by
70 •Journal of School Health •January 2025, Vol. 95, No. 1
©2024 The Author(s). Journal of School Health published by Wiley Periodicals LLC on behalf of American School Health Association.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium,
provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
the American Academy of Sleep Medicine.6Like
physical activity, teen sleep duration has been steadily
degrading in recent decades.7A body of research
has sought to untangle the relationships between
sleep and activity. Several have concluded that
increased physical activity correlates with longer sleep
in adolescents.8,9 A small randomized, short-term
experimental trial of sleep extension in adolescents
showed that longer sleep led to less sedentary time.10
With regards to the day-to-day relationship between
adolescents’ sleep and activity influencing each other,
a study where adolescents wore accelerometers for
a week revealed that more moderate or vigorous
physical activity led to earlier bedtimes, longer
duration and greater sleep efficiency in the subsequent
sleep period, but longer sleep periods predicted both
lower amounts of sedentary and moderate or vigorous
physical activity in the next day.11 Unfortunately,
there have not been many opportunities to examine
what happens to activity patterns in a larger,
population-based group of adolescents over longer
periods of time (years rather than days) when sleep
has been enhanced.
Healthful teen sleep is particularly vulnerable to
early school start times, as adolescent circadian shifts
drive teens to later sleep and wake times.12 Therefore,
school start time policy changes may be an effective
avenue to conduct such an examination. Delayed high
school start times (starting at 8:30 AM or later13) extend
sleep duration,14 and by this mechanism also may
alter activity patterns. However, sleep extension is
not the only pathway by which school start time
change could impact activity. For instance, parents
of children whose schools started early reported that
their children prioritize sleep over active transport
to school (walking or biking) and that they would
feel unsafe allowing their child to walk or bike to
school when it was dark outside.15 One concern voiced
by community members when a school district was
considering delaying a high school’s start time was
that the new schedule will be less conducive for the
school’s athletic program if schools run later and there
was less afternoon time free for organized sports.16
In the COMPASS study in Ontario, Canada, start
dResearch Assistant, (billm058@umn.edu), Division of Epidemiology and Community Health, University of Minnesota School of Public Health, Minneapolis, MN.
eSenior Research Fel low, (wahls001@umn.edu), Department of Organizati onal Leadership, Policy and Development, College of Education and Human Development, University of
Minnesota, Minneapolis, MN.
fProfessor, (nels5024@umn.edu), Division of Epidemiology and Community Health, University of Minnesota School of Public Health, Minneapolis, MN.
gProfessor, (widome@umn.edu), Divisionof Epidemiology and CommunityHealth, University of Minnesota School of PublicHealth, Minneapolis, MN.
Address corres pondence to: Rachel Widome, Profess or, (widome@umn.edu), Divisi on of Epidemiology and Community Health, 1300 Sout h Second Street, Suite 300, Minneapolis,
MN 55454.
This st udy is supported by funding from the National Institut es of Health’s (NIH) Euni ce Kennedy Shriver Nat ional Insti tute of Child Heal th and Human Development (NICHD) (R01
HD088176) and theNational Heart, Lung, and Blood Institute (R21 HL156218). The effort of K.T.J. was supported by AwardNumber T32HL150452(PI: D. Neumark-Sztainer) fromthe
National Heart, Lung, and Blood Institute (NHLBI). Additionally, the authors gratefully acknowledge support from the MinnesotaPopulation Center (P2C HD041023) funded
through a grant from NICHD. The authors would like to thank the adolescents participating in the START study, the districts that welcomed us to do researchi n their schools, the
START data collectors, and Bill Baker for his work to manage the data. Thank youto Kate Bauer for sharing your great ideas.
time changes appeared to have mixed effects on self-
reported physical activity, though it should be noted
that COMPASS was only able to examine very small
changes in start times, roughly 5-10 minutes.17
It is timely to ask questions about how high school
start time might influence activity as there has been a
push to delay start times given documented benefits to
teen sleep and wellbeing. Though many leading bodies
have recommended that high school not start earlier
than 8:30 AM (eg, American Academy of Pediatrics,
US Surgeon General),18 the vast majority of high
schools in the US still start early, with fewer than one
third of public secondary schools starting at 8:30 AM
or later.19 However, local communities and states are
considering and/or implementing policies that would
delay starts.20
The present study aims to characterize relationships
between delayed school start time policies and
adolescent physical activity, video game/computer
screen time, and extracurricular activity and organized
sports participation. The START study was a multi-
site evaluation of a natural experiment, assessing the
effects of a school start time policy change in high
schools in the Minneapolis, Minnesota metropolitan
area by comparing a large cohort of students in
previously early-starting schools that delayed start
times by 50 and 65 minutes, to students in schools that
maintained a consistent early start time.21 It has been
demonstrated previously that after the start time delay
was implemented, students attending policy change
schools in the START study had roughly 40 minutes
more sleep on school nights relative to students in
comparison schools.14
METHODS
Participants
Design. The START Study collected longitudinal
data from a large, grade-level cohort of students in
5 schools, 2 that had a district-initiated shift to a
later start time (‘‘policy change’’ schools) and 3 that
maintained a consistent, early start time (7:30 am)
(‘‘comparison’’ schools) over a 3-year period (Baseline,
Follow-up 1, and Follow-up 2). The 2 schools that
Journal of School Health •January 2025, Vol. 95, No. 1 •71
©2024 The Author(s). Journal of School Health published by Wiley Periodicals LLC on behalf of American School Health Association.
shifted to a later start time during the study began
at 7:30 AM and 7:45 AM in the 2015-2016 school year
(Baseline), and delayed their start times by 50 and
65 minutes, respectively, to 8:20 AM and 8:50 AM,in
the 2016-2017 and 2017-2018 school years (Follow-
up 1 and Follow-up 2).
Recruitment and data collection. All parents or
guardians of ninth graders in the participating high
schools were informed by letter of the START Study,
explaining the voluntary nature of the study and how
to opt out of study procedures. On measurement days,
participants were given information about the study,
that it was voluntary, and then were able to assent
to participation. Surveys were collected from 2134
grade 9 students at Baseline, 1934 grade 10 students
at Follow-up 1, and 1878 grade 11 students at Follow-
up 2. Participants were linked across waves with a
randomly assigned study identifier. All participants
who completed a START survey at any time point
were eligible to be included in this study. All survey
collection took place between February 28 and April
24 of each study year.
Instrumentation
Exposure. The exposure, school start time policy, is
a binary variable identifying if the participant attended
a ‘‘policy change’’ school.
Outcomes: physical activity. We used a summary
physical activity index modified from a validated
method published in 1985 by Godin and Shephard
to assess the mean physical activity of a community
before and after a community level health or
physical fitness intervention.22,23 The physical activity
questionnaire included as part of each START survey
consists of 3 questions about the number of hours
spent on ‘‘Strenuous exercise (heart beats rapidly),’’
‘‘Moderate exercise (not exhausting),’’ and ‘‘Mild
exercise (little effort),’’ in the past 7 days. Participants
selected the range of hours that included their effort
(eg, ‘‘2-4 hours a week’’). We assigned the midpoint
of that range (eg, ‘‘2-4 hours a week’’ =3 hours) as
the outcome for that level of exercise. We top-coded
the response ‘‘6+hours a week’’ with a value of 8.
We multiplied the number of hours of each level by
a scalar, corresponding to their expected metabolic
equivalent of task (MET; 9 ×strenuous, 5 ×moderate,
and 3 ×mild). The total score was the sum of MET
hours from each level of physical activity.
Outcomes: screen time. Electronic screen time was
assessed with a single question derived from the Youth
Risk Behavior Surveillance System (YRBSS).24 START
participants were asked, ‘‘On an average school day,
how many hours do you play video or computer
games, or use a computer for something that is not
schoolwork? Count time spent on things such as Xbox,
PlayStation, an iPad or other tablet, a smartphone,
YouTube, Facebook, or other social networking tools,
and the internet.’’ Participants selected from the
following answer choices: ‘‘none,’’ ‘‘less than 1 hour,’’
‘‘1 hour,’’ ‘‘2 hours,’’ ‘‘3 hours,’’ ‘‘4 hours,’’ and ‘‘5 or
more hours.’’ We converted the responses to linear
hours, using a value of 0.5 for ‘‘less than 1 hour’’ and
a value of 5 for ‘‘5 or more hours.’’
Outcomes: organized sports and extracurricular
activity participation. Our measure of sports partic-
ipation was derived from a question designed to assess
the timing of sports activities. Participants were asked,
‘‘during the last week, when did you participate in
organized sports or physical activities?’’ and selected all
that applied from the following options: ‘‘In the morn-
ing, before school,’’ ‘‘in the afternoon, after school,’’
‘‘during the school day,’’ ‘‘in the evening on days
that I had school,’’ ‘‘on the weekend,’’ and ‘‘I did not
participate in organized sports or physical activities
in the last week.’’ We also assessed, using the same
method, the effect of school start time on participation
in ‘‘organized extracurricular activities.’’
Potential confounders. All weighted models were
adjusted via standardization, described in detail below,
for demographic and economic variables that are estab-
lished risk factors for physical inactivity, and could
be associated with school start time delay through
nonrandom distribution in early and late-starting
schools. Prior studies have indicated that prevalence
of adolescent physical activity varies by age, gender,
body mass index (BMI), race and ethnicity, and family
affluence.25 Due to the single age cohort included in
the START study, adjustment for age is not necessary.
Confounding variables as self-reported in the START
survey are self-reported biological sex (male, female),
BMI (continuous; calculated from self-reported height
and weight), eligibility for free/reduced-price meals
(yes, no, don’t know), parents’ highest educational
attainment (at least 1 parent completed college
or higher education, neither parent completed col-
lege), Hispanic ethnicity (Hispanic, non-Hispanic)
and race (white alone, Asian/black/Native Ameri-
can/multiracial/other people of color, used as a general
marker of exposure to racism). We treated the first
reported value of each of these variables as a time-
invariant confounder. We also considered baseline
(pre-policy change) physical activity, screen time, and
sports and extracurricular participation, to be con-
founding variables, because they differed by school,
which determined school start time, and are strong
predictors of each outcome at follow-up.
Data Analysis
Missing data. Accounting for differential censoring
was particularly important in the START Study, as
participants in early starting schools had a higher rate
72 •Journal of School Health •January 2025, Vol. 95, No. 1
©2024 The Author(s). Journal of School Health published by Wiley Periodicals LLC on behalf of American School Health Association.
of attrition, which was also associated with baseline
levels of some outcomes.
Nearly half of early starting students in the lowest
quartile of physical activity at baseline were censored
during START follow-up, compared to just one
fourth of students in the highest quartile of physical
activity in late-starting schools. Similar differential
attrition was observed for screen time use and sports
and extracurricular activity participation at baseline.
Failing to account for this differential attrition could
create selection bias, preventing an unbiased estimate
of the treatment effect.26
We used multiple imputation27 using fully
conditional specification in SAS version 9.4 (SAS
Institute, Cary, NC) to account for the censored
observations that resulted from differential attri-
tion. The discriminant/multinomial function was
used to impute school (5 levels), self-reported
biological sex (male/female/prefer not to answer),
eligibility for free/reduced-price meals (no/yes/don’t
know), each parent’s educational attainment (<high
school/high school/some college/college/advanced
degree/don’t know/no second parent or guardian),
and race category (American Indian or Alaska
Native/Asian/black, African or African Ameri-
can/Native Hawaiian or Pacific Islander/white or
Caucasian/multiracial/unknown or not reported). The
logistic function was used to impute responses to each
of the 3 exercise questions (none/less than 1/2 hour a
week/1/2-2 hours a week/2-4 hours a week/4-6 hours
a week/6+hours a week), organized sports partici-
pation (no/yes), extracurricular activity participation
(no/yes), and Hispanic ethnicity (Hispanic/non-
Hispanic), and each of the 6 depressed mood scale
items (Not at all/Somewhat/Very Much). Linear
regression with predictive mean matching was used
to impute continuous hours of electronic screen time.
Linear regression was used to impute the number of
hours of self-reported weekday sleep. Sleep duration
and depressed mood questions were included as aux-
iliary variables in the multiple imputation models, and
are not used elsewhere in this analysis. Each outcome
variable was missing from 12% to 19% of participants
at Baseline, 20% to 23% of participants at Follow-up
1, and 22% to 25% at Follow-up 2. At all time points,
most missing responses were due to non-participation
in the survey, rather than selective non-response.
Inverse probability of treatment weight estimation.
To control for confounding by baseline differences,
we applied an inverse probability of treatment
weighting (IPTW) approach.28 Eligible participants
were weighted using stabilized inverse probability
of treatment weights to reflect the demographic
characteristics and baseline physical activity, screen
time, and sports and activity participation rate of the
total sample.
Main analyses. The causal effect of later school
start times on each outcome at follow-up can be
estimated using a difference-in-differences estimator,
in which the changes in each outcome observed in the
comparison schools are used to estimate the changes
in each outcome that would have been observed in
the late-start adopting schools had they not delayed
their start times prior to Follow-up 1. For each model,
we regressed the continuous or binomial outcome
on time (Baseline, Follow-up 1, Follow-up 2), school
(4 dummy variables, omitting the largest comparison
school), and delayed school start time at Follow-up 1
and Follow-up 2 (delayed versus early start at each
time point):
Yijt =β0+β1Timet+β2Schoolj+β3Delayed_Time
2jt +β4Delayed_Time 3jt +itj
where Yijt is the outcome of interest for person i
in school jat time t,β0is the mean outcome in
the largest comparison school at the reference time
period (Baseline), β1is the change in the outcome in
comparison schools at time t,andβ2is the fixed effect
for school j.β3and β4are the primary hypothesis tests,
the difference in differences for delayed-start adopting
schools, compared to comparison schools, at Follow-up
1 and Follow-up 2, respectively. ijt is the unexplained
residual variance for person iin school jat time t.
All models were fitted with IP-weighted generalized
estimating equation (GEE) regression models, with
working unstructured correlation matrix. GEE models
provide a ‘‘robust’’ Wald confidence interval that
guarantees coverage probability of at least 95%
when weighted regression is used. Time-by-condition
marginal means, and model parameter estimates, from
each of the 20 imputed data sets were pooled to
generate valid statistical inferences. The differences-in-
differences from these models estimate the association
between delayed school start time and each outcome
in the units of that outcome (MET hours of exercise,
hours and minutes of electronic screen time, and
percentage points of participation in organized sports
and extracurricular activities).
RESULTS
START surveys were completed by 2419 individual
participants at one or more time points. At Baseline,
2084 completed the physical activity questionnaire,
with 1914 completed at Follow-up 1 and 1864
completed at Follow-up 2. Nine participants did not
complete the physical activity questionnaire at any
time point. There were 1414 participants in delayed-
start schools and 1005 in comparison schools. We
further excluded 130 students who did not provide a
valid response at any time point for any covariates or
who indicated ‘‘prefer not to answer’’ for gender, for
a final analytic sample size of 2289.
Journal of School Health •January 2025, Vol. 95, No. 1 •73
©2024 The Author(s). Journal of School Health published by Wiley Periodicals LLC on behalf of American School Health Association.
Table 1. Baseline Self-Reported Demographic Characteristics of START Study Participants by Condition Characteristic
Study Group, No. (%)*
Characteristic Delayed-Start Schools (n =1414) Comparison Schools (n =1005) p-Value
Biological sex .68
Male 713 (50.4) 525 (52.2)
Female 692 (48.9) 474 (47.2)
Hispanic or Latino ethni city 56 (4.0) 37 (3.7) .64
Race <.001
Native American 15 (1.1) 9 (0.9)
Asian 131 (9.3) 23 (2.3)
Hawaiian/Pacific I slander 1 (0.1) 4 (0.4)
Black 69 (4.9) 25 (2.5)
White 1046 (74.0) 869 (86.5)
Multiracial 101 (7.1) 51 (5.1)
Other/unknown 51 (3.6) 24 (2.4)
Free or reduced-price eligi ble .002
No 971 (68.7) 643 (64.0)
Yes 201 (14.2) 134 (13.3)
Do not know 232 (16.4) 210 (20.9)
≥1 Parent completed college 1105 (78.2) 666 (66.3) <.001
∗Percentages may not total 100 owing to missing responses. Data are from the START Study, spring 2016 through spring 2018.
The sample was approximately 50% male, 95%
non-Hispanic, and 80% white. Fourteen percent
reported being eligible for free/reduced-price meals.
Nearly three-quarters of the students had at least
1 parent who completed college. Table 1shows
there were differences between students in policy
change schools versus those in comparison schools in
proportions of self-identified race groups and college
completion by at least 1 parent (Table 1).
Effect of School Start Time Delay on Physical Activity,
Screen Time, and Sports and Activity Participation
The weighted mean baseline physical activity level
was 60.8 MET hours per week in policy change schools,
and 61.4 MET hours per week in comparison schools,
and did not differ significantly by condition. Compared
to students in early starting comparison schools, the
changes in physical activity for students starting school
later were not significantly different at either Follow-
up1(β=−1.5, 95% confidence interval [CI]: −4.9,
1.9) or Follow-up 2 (β=1.5, 95% CI: −2.2, 5.2;
see Table 2). These differences are equivalent to one
half hour of mild activity, or 10 minutes of strenuous
activity, per week.
In both policy change schools and comparison
schools, students reported electronic screen use of
2 hours, 36minutes at Baseline. Over 2 years of follow-
up, screen time increased by 6 minutes per day in
both conditions, but the changes were not different at
either Follow-up 1 (β=0:03, 95% CI: −0:06, 0:12) or
Follow-up 2 (β=0:02, 95% CI: −0:06, 0:12).
School start time was also not associated with
differences in organized sports or extracurricular
activity participation during follow-up. Organized
sports participation declined by 19 percentage points
over 2 years in both delayed-start adopting schools and
comparison schools (difference-in-differences =0.00,
95% CI: −0.05, 0.06). In contrast, extracurricular
activity participation was stable in both conditions,
with no change in delayed-start adopting schools,
and a 1 percentage point increase in comparison
schools (difference-in-differences =0.00, 95% CI:
−0.05, 0.06).
DISCUSSION
The START Study provided an opportunity to
assess how school start time influenced adolescents’
activity behaviors. The extended sleep offered by later
school start times could have left adolescents less
tired, and more able to engage in physical activity.
However, extending sleep also reduced awake time
by an average of 40 minutes per school day, which
could have displaced physical activity, and altered
school schedules could have interfered with sports
and extracurricular activity meeting times. Similarly,
students with less pressing need to wake up early in
the morning could have used evening time to extend
their screen usage. However, in our analysis we saw
no evidence that the start time change had an impact
on these outcomes.
While this means that we did not find evidence
supporting that a later start time promoted more
healthful activity behaviors, we also did not find
evidence that later school start times were harmful
to these domains. This provides some answers to the
persistent claims that delayed school start times could
interfere with organized youth sports. Understanding
the impact of start time delay on the timing of sports
and activity participation was beyond the scope of
74 •Journal of School Health •January 2025, Vol. 95, No. 1
©2024 The Author(s). Journal of School Health published by Wiley Periodicals LLC on behalf of American School Health Association.
Table 2. Weighted*Difference-in-Differences Analysis of Changes in Physical Activity, Screen Time, Sports Participation, and Extracurricular Activity Participation Among High
School Students (Grade 9 at Baseline to Grade 11 at Follow-up 2) Before and After a 50- to 65-Minute Delay in School Start Times; START 2016-2018
Delayed-Start Schools Comparison Schools Difference-in-Differences Analysis
Variable
Baseline
Mean (95% CI)
Follow-
Up 1
Follow-
Up 2 Baseline
Follow-
Up 1
Follow-
Up 2
Baseline to
Follow-Up 1 p
Baseline to
Follow-Up 2 p
Physical activity (MET
hours per week)
60.8 (58.7, 62.8) 52.1 (50, 54.1) 50.6 (48.4, 52.8) 61.4 (59.1, 63.7) 54.2 (51.6, 56.8) 49.7 (47, 52.4) −1.47 (−4.9, 1.9) 0.39 1.54 (−2.2, 5.2) 0.41
Electronic screen time
(h:min per day)
2 h:36 min ( 2:29, 2: 42) 2 h:36 min ( 2:28, 2: 42) 2 h:42 min ( 2:35, 2: 48) 2 h:36 min ( 2:31, 2: 42) 2 h:36 min (2:26, 2:42) 2 h:42 min (2:33, 2:48) 0 h:03 min ( −0:06, 0: 12) 0.60 0 h: 02 min (−0:06, 0:12) 0.72
Organized s ports parti c-
ipation
0.74 (0.71, 0.76) 0.63 (0.6, 0.66) 0.55 (0.52, 0.58) 0.72 (0.69, 0.75) 0.61 (0.58, 0.65) 0.53 (0. 48, 0.57) 0.00 (−0.05, 0.05) 0.92 0.01 (−0.05, 0.06) 0.82
Extracurricular activity
participation
0.50 (0.47, 0.53) 0.49 (0.46, 0.52) 0.51 (0.48, 0.54) 0.49 (0. 46, 0.53) 0.51 (0.47, 0.55) 0.49 (0.46, 0. 53) −0.03 (−0.08, 0.03) 0.38 0.00 (−0.05, 0.06) 0.92
Abbreviations: CI, confidence interval; H:MM, hours: minutes; MET, metabolic equivalent of task.
∗Data are from the START Study, spring 2016 through spring 2018. Baseline occurred in spring 2016 (9th grade), follow-up 1, spring 2017 (10th grade), and follow-up 2, spring 2018 (11th grade). All estimates from generalized linear
mixed models. Inverse probability of treatment weights used to weight each condition to full sample distribution of baseline physical activity score, hours of screen time, and organized sports and extracurricular activities participation,
body mass index, biological sex, free/reduced-price meal eligibility, parent’s educational attainment, race and ethnicity.
Journal of School Health •January 2025, Vol. 95, No. 1 •75
©2024 The Author(s). Journal of School Health published by Wiley Periodicals LLC on behalf of American School Health Association.
this study, but could be a fruitful avenue for future
research. Our results were consistent with what was
found in COMPASS, where start time change appeared
to have no impact on screen time and an ambiguous
connection with moderate and vigorous physical
activity.17 As previously mentioned COMPASS had
the notable limitation that participating schools start
times changed by only 5-10 minutes.17
Some limitations may affect the validity of these
findings. Relying on self-reported measures of physical
activity and electronic screen time certainly results in
measurement error (and its degree is unknown for
measures that have not been validated). This can
reduce power to detect a true effect or lead to a
mischaracterization of associations. Additionally, some
question wording may also have been ambiguous
(such as asking about participation in ‘‘organized
sports or physical activities,’’ which could reasonably
be interpreted as either including or excluding physical
education classes). Other questions, notably the
electronic screen time question, had limited response
categories, which may not have been sensitive to
smaller changes in device use.
Other, contextual, factors may affect generalizability
of these findings. The schools in the START Study
were suburban and exurban/rural. While one potential
benefit of later school start times could be enabling
walking or bicycling to schools, the communities these
schools were situated in were not well suited to active
transportation, and therefore early start times may
have been only one of several barriers to increasing
physical activity in this way. Delaying start times in an
urban context may better facilitate these behaviors.
Some school-level variables that we could not incor-
porate into the model may also have influenced
change over time, separately from school start times.
For example, each school district has different phys-
ical education requirements. Delayed-start adopting
schools required somewhat more physical education
credits, as a percent of required credits, but it was
not possible to add that school-level variable while
also including a fixed effect for school. It is likely that
school districts that adopted one policy prioritizing the
health of students may also differ in other, unmea-
sured, ways from schools that did not adopt such a
policy. Similarly, it would have been preferable to
adjust for the time-varying month of survey adminis-
tration, which may be associated with organized sports
participation due to different sports seasons.
IMPLICATIONS FOR SCHOOL HEALTH POLICY, PRACTICE, AND
EQUITY
Communities interested in promoting sleep by
delaying start times may do so knowing that there are
unlikely to be adverse effects on adolescent physical
activity, electronic screen time, or organized sports and
activity participation.
Conclusions
Despite these limitations, this study is the first ever
comprehensive evaluation of the effects of school start
time delay on these outcomes. Over 2 years of follow-
up, no changes emerged to suggest that later school
start times either interfered with, or promoted, any
of these outcomes. Starting high school later is a
policy that has been demonstrated to promote health
and wellbeing, with and does not appear to adversely
impact physical activity.
Human Subjects Approval Statement
All study procedures were reviewed and approved
by the University of Minnesota Institutional Review
Board (IRB) and the school districts’ research review
panels.
Conflict of Interest
The authors have no conflicts of interest to disclose.
REFERENCES
1. US Department of Health and Human Services. Physical
Activity Guidelines for Americans. 2nd ed. Washington, DC: US
Department of Health and Human Services; 2018.
2. Kann L, McManus T, Harris WA, et al. Youth risk behavior
surveillance— United States, 2017. MMWR Surveill Summ.
2018;67(8):1-114. https://doi.org/10.15585/mmwr.ss6708a1.
3. Brown DMY, Summerville B, Fairclough SJ, Mielke GI,
Tyler R. Associations between intersecting sociodemographic
characteristics and device-measured physical activity among
children and adolescents living in the United States. J Phys
Act Health. 2024;21:384-393. https://doi.org/10.1123/jpah.2023
-0360.
4. Loux T, Matusik M, Hamzic A. Trends in U.S. adolescent
physical activity and obesity: a 20-year age-period-cohort
analysis. Pediatr Obes. 2023;18(3):e12996. https://doi.org/10
.1111/ijpo.12996.
5. Ryu S, Kim H, Kang M, Pedisic Z, Loprinzi PD. Secular trends in
sedentary behavior among high school students in the United
States, 2003 to 2015. Am J Health Promot. 2019;33(8):1174-
1181. https://doi.org/10.1177/0890117119854043.
6. Paruthi S, Brooks LJ, D’Ambrosio C, et al. Consensus state-
ment of the American Academy of sleep medicine on the
recommended amount of sleep for healthy children: method-
ology and discussion. J Clin Sleep Med. 2016;12(11):1549-1561.
https://doi.org/10.5664/jcsm.6288.
7. Keyes KM, Maslowsky J, Hamilton A, Schulenberg J. The
great sleep recession: changes in sleep duration among
US adolescents, 1991-2012. Pediatrics. 2015;135(3):460-468.
https://doi.org/10.1542/peds.2014-2707.
8. Ganz M, Jacobs M, Alessandro C, et al. Physical activity
and sleeping duration among adolescents in the US. Cureus.
2022;14(9):e29669. https://doi.org/10.7759/cureus.29669.
9. Desjardins S, Tanguay-Labont ´
e M. The effects of physical
activity on sleep among adolescents and adults: a narrative
review. JSDR. 2018;1(2):42-59. https://doi.org/10.14302/issn
.2574-4518.jsdr-18-2065.
76 •Journal of School Health •January 2025, Vol. 95, No. 1
©2024 The Author(s). Journal of School Health published by Wiley Periodicals LLC on behalf of American School Health Association.
10. Van Dyk TR, Krietsch KN, Saelens BE, Whitacre C, McAlister S,
Beebe DW. Inducing more sleep on school nights reduces
sedentary behavior without affecting physical activity in short-
sleeping adolescents. Sleep Med. 2018;47:7-10. https://doi.org
/10.1016/j.sleep.2018.03.007.
11. Master L, Nye RT, Lee S, et al. Bidirectional, daily temporal
associations between sleep and physical activity in adolescents.
Sci Rep. 2019;9(1):7732. https://doi.org/10.1038/s41598-019
-44059-9.
12. Owens J, Au R, Carskadon M, et al. Insufficient sleep in
adolescents and young adults: an update on causes and
consequences. Pediatrics. 2014;134(3):e921-e932. https://doi
.org/10.1542/peds.2014-1696.
13. Adolescent Sleep Working Group, Committee on Adolescence,
Council on School Health et al. School start times for
adolescents. Pediatrics. 2014;134(3):642-649. https://doi.org/10
.1542/peds.2014-1697.
14. Widome R, Berger AT, Iber C, et al. Association of
Delaying School Start Time with Sleep Duration, timing, and
quality among adolescents. JAMA Pediatr. 2020;174(7):697-704.
https://doi.org/10.1001/jamapediatrics.2020.0344.
15. Ahlport KN, Linnan L, Vaughn A, Evenson KR, Ward
DS. Barriers to and facilitators of walking and bicycling to
school: formative results from the non-motorized travel study.
Health Educ Behav. 2008;35(2):221-244. https://doi.org/10.1177
/1090198106288794.
16. Kubow PK, Wahlstrom KL, Bemis AE. Starting time and school
life: reflections from educators and students. Phi Delta Kappan.
1999;80(5):366-371.
17. Patte KA, Qian W, Cole AG, et al. School start time changes in
the COMPASS study: associations with youth sleep duration,
physical activity, and screen time. Sleep Med. 2019;56:16-22.
https://doi.org/10.1016/j.sleep.2018.09.020.
18. Start School Later. Key Position Statements; 2024. Available
at: https://www.startschoollater.net/key-position-statements
.html.
19. U.S. Department of Education, National Center for Education
Statistics. National Teacher and Principal Survey (NTPS),
‘‘Public School Data File,’’ 2020-21; Available at: https://nces
.ed.gov/surveys/ntps/estable/table/ntps/ntps2021_srt04_s1s.
20. Start School Later. Legislation; Available at: https://www
.startschoollater.net/legislation.html.
21. Widome R, Wahlstrom KL, Laska MN, et al. The START study:
an evaluation to study the impact of a natural experiment in
high school START times on adolescent weight and related
behaviors. Obs Stud. 2020;6:66-86.
22. Godin G, Shephard RJ. A simple method to assess exercise
behavior in the community. Can J Appl Sport Sci. 1985;10(3):
141-146.
23. Sallis JF, Buono MJ, Roby JJ, Micale FG, Nelson JA.
Seven-day recall and other physical activity self-reports
in children and adolescents. Med Sci Sports Exerc. 1993;
25(1):99-108. https://doi.org/10.1249/00005768-199301000
-00014.
24. 2017 State and Local Youth Risk Behavior Survey; Centers for
Disease Control and Prevention. Available at: https://www.cdc
.gov/healthyyouth/data/yrbs/pdf/2017/2017_yrbs_standard
_hs_questionnaire.pdf.
25. Iannotti RJ, Wang J. Trends in physical activity, sedentary
behavior, diet, and BMI among US adolescents, 2001-2009.
Pediatrics. 2013;132(4):606-614. https://doi.org/10.1542/peds
.2013-1488.
26. Hern´
an MA, Robins JM. Causal Inference: What if . Boca Raton:
Chapman & Hall/CRC; 2024.
27. Li P, Stuart EA, Allison DB. Multiple imputation: a flexible
tool for handling missing data. JAMA. 2015;314(18):1966-1967.
https://doi.org/10.1001/jama.2015.15281.
28. Robins JM, Hern ´
an MA, Brumback B. Marginal struc-
tural models and causal inference in epidemiology. Epidemi-
ology. 2000;11(5):550-560. https://doi.org/10.1097/00001648
-200009000-00011.
Journal of School Health •January2025,Vol.95,No.1 •77
©2024 The Author(s). Journal of School Health published by Wiley Periodicals LLC on behalf of American School Health Association.
