Using Ecological Momentary Assessment Methods in Mind-Body Research

1
Using Ecological Momentary Assessment
Methods in Mind-Body Research
Thomas W. Kamarck, Ph.D. University of
Pittsburgh Pittsburgh Mind-Body Center July 10,
2006
2
Collaborators Lori Arnold, M.S., Mary Witzig,
George Haff Saul Shiffman, Ph.D. Karen Matthews,
Ph.D. Dan Buysse, M.D. Wesley Thompson,
Ph.D. Vicki Helgeson, Ph.D. Rebecca Thurston,
Ph.D.
3
FORMAT OF TODAYS SESSION 1. For
Pre-Contemplators Pros and Cons of
EMA --History --Why EMA considered as
alternative to traditional assessment --What
questions are best asked with these
methods --How information gleaned from these
methods may be different from that obtained from
alternative approaches
4
FORMAT OF TODAYS SESSION 2. For
Contemplators Preparing to take the EMA
plunge --Design alternatives --Options and
opportunities
5
FORMAT OF TODAYS SESSION 3. For Action
Phase Implementation of EMA methods --
Choosing Equipment -- Designing
Instruments -- Training Participants --
Enhancing Compliance -- Managing Data
6

• FORMAT OF TODAYS SESSION
• INTRODUCTION
• Kamarck 900-1015
• BREAK 1015-1030
• 2. APPLICATIONS
• Matthews, Buysee, Thompson 1030-100
• Helgeson, Thurston, Kamarck
• LUNCH 100-145
• IMPLEMENTATION
• Witzig Haff 145-300

7
ECOLOGICAL MOMENTARY ASSESSMENT
IN THE NATURAL ENVIRONMENT
IN REAL TIME
DATA COLLECTION BY STUDY PARTICIPANTS
Stone Shiffman (1994). Annals Beh Med, 16,
199-202.
EXPERIENCE SAMPLING Csikszentmihalyi et al.,
1977 AMBULATORY ASSESSMENT Fahrenberg Myrtek,
2001
8
EMA PRECURSORS

• Time budget analysis
• (Thorndike et al., 1937)
• Analysis of behavioral settings
• (Barker Wright, 1951 One boys day)
• Self-monitoring and behavior change
• (McFall, 1977)
• Experience sampling
• Czikszentmihalyi (1990) Flow

9
EMA APPLICATIONS

• Social Interactions
• (Reis Wheeler, 1991)
• Psychopathology
• (deVries, 1992)
• Psychophysiological Assessment
• (Fahrenberg Myrtek, 2001)

10
THE SCIENTIFIC LITERATURE REFLECTS A GROWING
INTEREST IN THESE AREAS IN RECENT YEARS
OF MEDLINE REFERENCES
CITED ARTICLES USING KEYWORDS
Ecological Momentary Assessment,
Experience Sampling,
Ambulatory Assessment.
11
PROS AND CONS OF EMA

• Traditional assessment approaches
• have significant limitations.
• 2. EMA approaches tell a different story (perhaps
  more valid?) than traditional approaches.
• 3. EMA approaches may provide some advantages for
  addressing certain types of questions.
• 4. EMA approaches may not be advantageous in all
  circumstances.
• 5. EMA approaches may have some limitations.

12

1. Traditional assessment approaches have
   significant limitations.

• AUTOBIOGRAPHICAL MEMORY
• Not accurate in recalling details of daily
  lives.
• Errors are systematic, not random.
• Memory as process of reconstruction rather than
  retrieval.

13

1. Traditional assessment approaches have
   significant limitations.

• HEURISTICS AFFECTING RECOLLECTION OF RECENT
  EVENTS
• (Bradburn, Rips Shevell, 1987, Science, 236,
  157-161).
• Availability more salient events given
  greater weight.
• Recency more recent events have
  disproportionate impact.
• State biases effects of present mood.
• Effort after meaning memory adjusted to fit
  present circumstances.

14
2. EMA approaches tell a different story
(perhaps more valid?) than traditional
approaches.

1. RELAPSE PREVENTION MODEL
2. DAILY COPING
3. AMBULATORY VS. CLINIC BP
4. JOB DEMAND

15
2. EMA approaches tell a different story
(perhaps more valid?) than traditional
approaches.

• RELAPSE PREVENTION MODEL
• Transgressions (slips) following efforts at
  smoking cessation are common.
• Behavioral and affective response to slips
  predict subsequent prognosis. Coping efforts.
• Empirical work Retrospective reports.
• Coping behaviors following slip are associated
  with reduced risk for relapse.

16
2. EMA approaches tell a different story
(perhaps more valid?) than traditional
approaches.
a. RELAPSE PREVENTION MODEL Shiffman et al.
(1996) JCCP, 5, 993-1002. 133 smoking cessation
program participants Electronic diaries in weeks
following cessation. Reports of coping related to
subsequent lapse during same day, but not to
ultimate relapse status.
17
2. EMA approaches tell a different story
(perhaps more valid?) than traditional
approaches.
b. DAILY COPING Stone et al. (1998) JPSP, 74,
1670-1680. Compared reports of coping using EMA
vs. retrospective measures. 100 Ss with marital
or job stress EMA interviews every 40 mins over 2
days.
18
2. EMA approaches tell a different story
(perhaps more valid?) than traditional
approaches.
b. DAILY COPING Items endorsed in EMA interviews
were not endorsed in retrospective report 28 of
time on average. Items endorsed in retrospective
report not endorsed in EMA interviews 31 of the
time on average. Retrospective reports invoke
different types of cognitive processes than
momentary accounts.
19
2. EMA approaches tell a different story
(perhaps more valid?) than traditional
approaches.
c. AMBULATORY VS. CLINIC BP ABP measures are
better predictors of CVD than are clinic BP
measures. More representative, but also
measurement advantages. In 216 healthy adults, 4
automated BPs over 2.5 hours in laboratory and in
field. Compared these two sets of measures in
terms of association with carotid IMT.
20
2. EMA approaches tell a different story
(perhaps more valid?) than traditional
approaches.
c. AMBULATORY VS. CLINIC BP Kamarck et al. (2002)
J Hypertension, 20, 1535-41.
21
2. EMA approaches tell a different story
(perhaps more valid?) than traditional
approaches.
c. JOB DEMAND Kamarck et al. (2004) Health
Psychology, 23, 4-32. Perceived demand and
control using repeated EMA assessments over 6
day period in 330 adults. Demand, control also
assessed using global retrospective reports
Karasek Job Content questionnaire.
22
2. EMA approaches tell a different story
(perhaps more valid?) than traditional
approaches.
c. JOB DEMAND
n 152 Questionnaire Demand
Control Diary Task Demand r
.53 Decisional Control r .31
p lt .0001 after adjustment for age, sex and
education.
23
2. EMA approaches tell a different story
(perhaps more valid?) than traditional
approaches.
b.02, F (1, 328) 8.44, r2 .02, p .004
Kamarck et al. Health Psychology 2004
24
3. EMA approaches may provide some advantages
for addressing certain types of questions.
a. QUESTIONS ABOUT MEAN LEVELS OF CHARACTERISTIC,
WHEN CONTINUOUS BUT FLUCTUATING (e.g., mood).
b. QUESTIONS ABOUT FREQUENCY OR DURATION OF
DISCRETE EVENTS, WHEN FREQUENT AND NOT SALIENT
(e.g., social interactions).
c. QUESTIONS ABOUT VARIABILITY, WITHIN-PERSON
ASSOCIATIONS,OR TIME COURSE (e.g., stress and
BG).
d. QUESTIONS ABOUT SETTING DIFFERENCES OR
EFFECTS OF ENVIRONMENTAL EVENTS ON BEHAVIOR
(e.g., triggers of relapse).
25
4. EMA approaches may not be advantageous in
all circumstances.

• WHEN CHARACTERISTIC IS NOT EXPECTED TO VARY
• ACROSS TIME OR SETTING (trait adjectives, e.g.,
  dependable).

b. WHEN EVENT OCCURS SO RARELY NOT LIKELY TO BE
CAPTURED WITH MONITORING (e.g., doctors
apppointments).
c. WHEN EVENT IS SO SALIENT, NOT LIKELY TO
BE MISSED (e.g., AICD event).
d. WHEN PERSON IS NOT LIKELY TO HAVE ACCESS TO
THE EVENT OR CHARACTERISTIC (e.g., anger
inhibition).
26
4. EMA approaches may have some limitations.

1. CORRELATION VS. CAUSATION

NEED TO CONSIDER THIRD FACTORS (e.g., BP at work)
b. COMPLIANCE ISSUES
c. REACTIVITY OF ASSESSMENT
d. SAMPLE REPRESENTATIVENESS
e. LIMITATIONS OF SELF-REPORT
27
DESIGN FEATURES

• Time-based vs. event-based.
• Fixed vs. random sampling.
• 3. Sampling frequency, sampling coverage.
• 4. Combination assessment strategies.
• 5. Reporting platform (electronic vs. paper).
• 6. Factors affecting subject burden.
• 7. Factors affecting compliance.
• 8. Reactivity of assessment.

28
1. Time-based vs. event-based.
Continuous time-based approaches Discrete
event-based approaches
Advantages of event-based Can capture a high
proportion of events Disadvantages of
event-based Much is left up to the
subject Disincentive to be thorough Behavioral
samples are not representative cannot
generalize to person-level characteristics
29
2. Fixed vs. random sampling.

• Fixed
• Can synchronize with other devices.
• Helpful for data analysis?
• Assessments more predictable.
• Random
• Ensures that samples are representative.
• Interviews less likely to be entrained to
  regular events, e.g., hourly appointments.

30
3. Sampling frequency, sampling coverage.
What frequency is optimal? Is end-of-day sampling
sufficient?
Parkinson et al., 1995, PSPB, 21, 331-339. 30
Sx, momentary (every 2 hrs) vs. End of
Day Within-S corrs over 2 week period Negative
Affect r .68 Positive Affect r .66 Evidence
for independent contribution of peak, recent, and
concurrent mood on end of day ratings.
31
4. Combination Assessment Strategies.
Environmental, cognitive, or emotional correlates
of behavioral events (e.g., relationship between
stress and relapse) Event-based sampling
(record stress level prior to smoking) not
sufficient what is the base rate of
stress? Combine with time-based random sampling
(what is average stress rating throughout the
day). CASE-CROSSOVER DESIGN Shiffman Waters
(2004). JCCP, 72, 192-201.
32
5. Reporting Platform (Electronic vs. Paper).
Options Paper-pencil diaries Watches,
pagers PDAs Interactive Voice Response
(IVRs) Cell Phones PC and web-based response
devices
33
5. Reporting Platform (Electronic vs. Paper).
Options Paper-pencil diaries Watches,
pagers PDAs Interactive Voice Response
(IVRs) Cell Phones PC and web-based response
devices
Most widely used
34
5. Reporting Platform (Electronic vs. Paper).
Advantages of PDAs 1. Time stamped responses
--Permits synchronization with
physiological data. 2. Auditory
prompts --Enhances timely responding. 3.
Electronic data reduces data entry errors. 4.
Programmed code permits greater flexibility
in item administration. -- Filter questions,
protocol changes.
35
5. Reporting Platform (Electronic vs. Paper).
Advantages of PDAs 5. Subject
preference Hufford (in press) 4 studies,
diverse samples Exposed to EDs vs.
paper Average of 75 preferred EDs
36
5. Reporting Platform (Electronic vs. Paper).
Advantages of PDAs Stone et al. (2002) BMJ, 324,
1193-1194. 80 chronic pain patients Instrumented
diary vs. Compliance-enhanced ED Pain records 3
X/day, 10, 4, 8 /- 15 mins.
37
5. Reporting Platform (Electronic vs. Paper).
Advantages of PDAs
According to paper diary records 90 of
assessments were compliant. According to
photosensors 11 were compliant.
38
6. Factors Affecting Subject Burden.

• Hufford (in press)
• Density of sampling.
• Length of assessments.
• c. User-interface.
• d. Complexity of assessments.
• e. Duration of monitoring period.
• f. Stability of reporting platform.

39
7. Factors Affecting Compliance.
Hufford Shields (2002). Applied Clinical
Trials, 11, 46-56.
76 peer reviewed publications using electronic
diaries. 44 report compliance rates. Of these,
most report gt 80 compliance.
40
7. Factors Affecting Compliance.

• Hufford Shiffman (2003) Disease Mngment and
  Health Outcomes, 11, 77-86.
• Build compliance into protocol.
• Subject training.
• c. User-interface.
• d. Drive the protocol.
• e. Guide subjects through assessments.
• f. Livability functions.
• g. Create sense of accountability.

41
7. Factors Affecting Compliance.
h. Participant support (continuity of care,
accessibility, flexible scheduling, reminder
cards, ongoing telephone contacts, etc.)
42
8. Reactivity of Assessment
Stone et al. (2003).Pain, 104, 343-351. 91
chronic pain pts. Monitored 2 weeks 4
conditions 3 x day, 6 x day, 12 x day, no
momentary 18 questions about pain and
activities
43
Stone et al. (2003). Pain, 104, 343-351.
44
8. Reactivity of Assessment
Within-subject comparison N 118, 16 prompts (45
min) ABP vs. 5 prompts (135 min)
Adjacent weekend days. Scales 1-4
16 5 p value Negative Affect 1.4 1.4 .13 Aro
usal 3.2 3.1 .002 Demand 1.5 1.5 .99 Control 2.
9 3.0 .16 Social Conflict 1.2 1.2 .92
45
8. Reactivity of Assessment
Kamarck et al (unpublished) Within-subject
comparison N 118, 16 prompts (45 min) ABP vs.
5 prompts (135 min) Adjacent weekend days.
Scales 1-4
16 5 p value Negative Affect 1.4 1.4 .13 Aro
usal 3.2 3.1 .002 Demand 1.5 1.5 .99 Control 2.
9 3.0 .16 Social Conflict 1.2 1.2 .92
46
REPORTING GUIDELINES
Stone Shiffman (2002). Annals Beh Med, 24,
236-243.

1. Sampling
2. Data Collection Procedures
3. Data Acquisition Interface
4. Compliance
5. Participant Training
6. Data Management
7. Data Analysis

47
RECOMMENDED ASSESSMENTS
1. Comparable end-of-day and global reports
2. Compliance data e.g., proportion of responses
completed, missed prompts, abandoned interviews,
delayed interviews
3. Subjective burden How much of a burden was it
to participate in the study? How willing would
you be to participate in a study like this
again? How much did participating in this study
interfere with your usual activities?
4. Previous computer use