p - reach

Genevieve F. Dunton, PhD, MPH
Yue Liao, MPH
Keito Kawabata, MPA
Stephen Intille, PhD
Methodological Limitations of
Epidemiological Instruments
Recall Instruments (e.g., surveys)
Memory errors and biases
Not completed in the environment in which the behavior occur
Difficulty capturing intraindividual variability
Observational Tools (e.g., SOPLAY)
Often limited to a single setting
Do not measure mood or subjective perceptions
Objective Devices (GPS or accelerometer)
Difficulty differentiating activity type and travel mode
Do not measure mood or subjective perceptions
mHealth Technologies
Mobile phone ownership is common (68% of adults
worldwide and 75% of US high school students)
Adopted across SES groups and in developing
countries
“Apps” can deliver real-time surveys
Smart phones have built-in accelerometer,
GPS, camera, and video technology
Able to synch with other ambulatory
sensors via bluetooth (e.g., heart rate,
asthma inhalers, air pollution, UV)
Riley, Rivera, Atienza, Nilsen, Allison & Mermelstein, Transl Behav Med, 2011.
.
Methodological Benefits of Ecological
Momentary Assessment (EMA)
Ecological
Momentary
Real-world environments & experience
Provides ecological validity
Real-time assessment & focus
Avoids recall bias
Assessment
Self-report
Repeated, intensive, longitudinal
Allows analysis of physiological/
psychological/behavioral processes
over time
(Stone & Shiffman, 1994)
Ecological Momentary
Assessment (EMA)
Can simultaneously measure:
1) Behavior (eating, watching TV)
2) Where (home, playground, trail, sidewalk)
3) With whom (alone, friends, siblings)
4) Perceived characteristics (safety, traffic)
5) Mood (positive affect, negative affect, stress)
6) Cognitive/motivational factors (self-efficacy)
EMA Sampling Schedules
Interval-contingent- information recorded according to
specific pre-set time frames (e.g., at 8am and 12noon
everyday)
Signal-contingent – information recorded when
prompted, often at random times throughout the day
Event-contingent - information recorded during or after
a pre-determined behavior
Context-contingent – information recorded when a
context or environment is sensed (GPS, heart rate, etc)
Example EMA Procedures
• Loaned mobile phone (HTC Shadow, T-Mobile)
• Monitoring occurred across 4 days (Sat.-Tues)
for each wave
• Signal-interval contingent hybrid sampling
Ecologi cal Mom entary As ses sm ent Pr om pting Sc hedule
D ay
S a turd ay
S u nd ay
M on da y
T ue sd a y
6:3 0a m
X
X
X
X
8 a m1 0a m
X
X
X
X
10 am 1 2p m
X
X
X
X
1 2p m
- 2p m
X
X
X
X
2p m 4p m
X
X
X
X
4p m 6p m
X
X
X
X
6 pm 8 pm
X
X
X
X
8p m10 pm
X
X
X
X
How can EMA Advance
Physical Activity Research?
Examine where and with whom physical activity occurs.
Examine differences in physical activity intensity and
duration across contexts.
Examine contextual differences in mood during
physical activity.
Examine psychological antecedents to and
consequences of physical activity bouts.
Examine mood variability and physical activity.
Examine interpersonal influences on activity across the
day.
Location of Outdoor Physical Activity
by Group
Percent of Outdoor Physical
Activity Reports
70
60
50
40
30
20
10
0
Park/Trail
Road
Parking Lot
Smartgrowth
Sidewalk
Other
Comparison
Dunton, G. F., Intille, S., Wolch, J., & Pentz, M. Investigating the impact of a smart growth community on the
contexts of children’s physical activity using Ecological Momentary Assessment, Health & Place, 18, 76-84, 2012
Physical Activity Level by Social Context
(30-min. before EMA prompt)
400
350
Steps
300
250
200
150
100
50
0
Family and
friends
Friends only
Family only
Alone
Dunton, G. F., Liao, Y., Intille, S., Wolch, J., & Pentz, M. (2011). Social and physical contextual influences on
children’s leisure-time physical activity: An Ecological Momentary Assessment study. Journal of Physical Activity
and Health, 8(Suppl 1), S103-S108.
.
Mood During Physical Activity by
Physical Context
Enjoyment
Positive Affect
3
Average Mood Rating
Average Mood Rating
3
2.5
2
1.5
1
0.5
0
2.5
2
1.5
1
0.5
0
Outdoors
Yard
Other
Home Someone
else's
house
Outdoors
Yard
Other
Home
Someone
else's
house
Dunton, G. F., Liao, Y., Intille, S., Wolch, J., & Pentz, M. (2011). Social and physical contextual influences on
children’s leisure-time physical activity: An Ecological Momentary Assessment study. Journal of Physical Activity
and Health, 8(Suppl 1), S103-S108.
Lagged Effects Models
7:45am
Phy. Act.
SE
PA
NA
Energy
Fatigue
Demand
Control
11:45am
Phy. Act.
SE
PA
NA
Energy
Fatigue
Demand
Control
3:45pm
7:45pm
Phy Act.
SE
PA
NA
Energy
Fatigue
Demand
Control
Phy Act.
SE
PA
NA
Energy
Fatigue
Demand
Control
Results of Multilevel Model Predicting MVPAT
Variable
Coeff. (SE)
p
Self-EfficacyT-1
0.08 (0.02)
<.001
Positive AffectT-1
0.06 (0.02)
.003
Negative AffectT-1
-0.11 (0.03)
<.001
EnergyT-1
0.04 (0.02)
.066
FatigueT-1
-0.02 (0.01)
.135
ControlT-1
0.05 (0.02)
.004
DemandT-1
-0.01 (0.01)
.978
Pos. Soc. Inter. T
0.17 (0.05)
.001
Prob. Soc. Inter. T
0.02 (0.06)
.692
Stressful EventT
-0.01 (0.06)
.875
Dunton, G. F., Atienza, A., Castro, C. M., & King, A. C. (2009). Using ecological momentary assessment to
examine antecedents and correlates of physical activity bouts in adults age 50+ years: A pilot study. Annals of
Behavioral Medicine, 38, 249-255.
Context-sensitive EMA with
Instrumented asthma inhaler
• Bluetooth-enabled
communication to smartphone
• EMA survey prompted 3-4
minutes after each asthma inhaler use
Current Activity By Type of
EMA Prompt
Random Rescue
ChiEMA
CS-EMA Square
Using Technology
47.8%
7.1%
8.38**
Sports/Exercise
19.5%
50.0%
6.59*
Eating/Drinking
12.4%
14.3%
0.41
Reading/Homework
7.1%
0.0%
1.06
Going Somewhere
5.3%
7.1%
0.08
Sleeping
8.8%
14.3%
0.43
**p < .10, *p < . 05
Is
mood variability (i.e., instability,
fluctuation) related to levels of physical
and sedentary activity?
Low Variability Case
High Variability Case
4
3
2
1
0
Time
10
8
6
4
2
0
8am
9am
10am
11am
12pm
1pm
2pm
3pm
4pm
5pm
6pm
7pm
8pm
9pm
10pm
Negative Affect
5
8am
9am
10am
11am
12pm
1pm
2pm
3pm
4pm
5pm
6pm
7pm
8pm
9pm
10pm
Negative Affect
6
Time
Location Model of Within-Person Mean (β = 0.001, t = 1.46, p > .05)
Scale Model of Within-Person Variability (τ = 0.002, t = 6.32, p < .001)
Day-Level Dyadic Models
Mother
Child
Obs 1
Obs 1
Obs 2
Obs 3
Day 1
(Avg.)
Day 1
(Avg.)
Obs 2
Obs 3
Obs 4
Obs 4
Obs 1
Obs 1
Obs 2
Obs 3
Day 2
(Avg.)
Day 2
(Avg.)
Obs 2
Obs 3
Obs 4
Obs 4
Obs 1
Obs 1
Obs 2
Obs 3
Obs 4
Day 3
(Avg.)
Day 3
(Avg.)
Obs 2
Obs 3
Obs 4
Table: Day-level
Correlations
Children’s Physical Activity
Children’s TV/Video Games
Children’s Fruit/Veg. Intake
Children’s Soda Intake
Children’s Junk Food Intake
Mother’s Average
Daily Stressors
r
-.34
-.06
-.16
.22
.06
Challenges and Limitations
Data
• Missing data
• Reactance
• Participant burden
• Costs
Acknowledgments
Active Living Research #RWJF 65837 (Dunton, PI)
ACS Mentored Research Scholar Grant 118283-MRSGT-10-012-01-
CPPB (Dunton, PI)
Southern California Environmental Health Science Center pilot grant
(# 5 P30 ES07048-16) (Dunton, PI)
USC Institute for Health Promotion and Disease Prevention
Research (Dunton, PI)
Marilyn Li, MD
Rob McConnell, PhD
Keito Kawabata, MS (Project Manager)
Yue Liao, MS (Research Assistant)
Eldin Dzubur, MS (Research Assistant)
Yifei Sun (Programming)
Cesar Aranguri, Alex Lau, Mark Lamm, Anuja Shah

Download Report