Maintaining High Ambulatory Activity Levels of Sedentary Adults with a Reinforcement Thinning Schedule
Leonardo F. Andrade, Danielle Barry, Mark D. Litt, and Nancy M. PetryUniversity of Connecticut School of Medicine
Abstract
Physical inactivity is a leading cause of mortality. Reinforcement interventions appear useful for
increasing activity and preventing adverse consequences of sedentary lifestyles. This study
evaluated a reinforcement thinning schedule for maintaining high activity levels. Sedentary adults
(n=77) were given pedometers and encouraged to walk ≥10,000 steps/day. Initially, all
participants earned rewards for each day they walked ≥10,000 steps. Subsequently, 61 participants
were randomized to a monitoring only condition or a monitoring plus reinforcement thinning
condition, in which frequencies of monitoring and reinforcing walking decreased over 12 weeks.
The mean ± SD percentage of participants in the monitoring plusreinforcement thinning condition
who met walking goals was 83% ± 24% versus. 55% ± 31% for participants in the monitoring
only condition, p < .001. Thus, this monitoring plusreinforcement thinning schedule maintained
high rates of walking when it was in effect; however, groups did not differ at a 24-week follow-up.
Monitoring plus reinforcement thinning schedules, nevertheless, hold potential to extend benefits
of reinforcement interventions at low costs.
Keywords
contingency management; reinforcement schedule; walking; sedentary adults
Physical inactivity is now the fourth leading risk factor for mortality worldwide (World
Health Organization, 2010). The American College of Sports and Medicine recommends 30
minutes or more of moderate intensity cardiorespiratory exercise at least 5 days per week
(Garber et al., 2011), and this level of activity can help prevent cardiovascular diseases,
Type 2 diabetes, and obesity (Boone-Heinonen, Evenson, Taber, & Gordon-Larsen, 2009;
Haskell et al., 2007; Hu, Li, Colditz, Willett, & Manson, 2003; Hu et al., 1999). One form of
exercise that is convenient and widely accessible is walking. Walking a minimum of 10,000
steps per day is usually equivalent to meeting the prescribed moderate intensity exercise
levels (Le-Masurier, Sidman, & Corbin, 2003) and is recommended in public-health activity
guidelines (Tudor-Locke & Bassett, 2004; Tudor-Locke et al., 2011; Tudor-Locke, Hatano,
Pangrazi, & Kang, 2008). Despite the benefits of walking, it is estimated that less than 5%
Correspondence should be sent to Nancy M. Petry, Calhoun Cardiology Center, University of Connecticut School of Medicine, 263 Farmington Avenue, Farmington, CT 06030-3944. Phone: 860-679-2593; Fax: 860-679-1312, [email protected] Barry is now at the Edith Nourse Rogers VA Hospital.
HHS Public AccessAuthor manuscriptJ Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Published in final edited form as:J Appl Behav Anal. 2014 ; 47(3): 523–536. doi:10.1002/jaba.147.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
of the US adult population engages in the recommended level of physical activity (Trojano
et al., 2008).
Many interventions that promote physical activity use pedometers (see Tudor-Locke et al.,
2011 for a review). Pedometers are light, unobtrusive, and relatively inexpensive monitors
that continuously measure the number of steps taken throughout the day(s). Recently, some
behavior analytic research has used pedometers to increase physical activity in adults and
children (see Van Camp & Hayes, 2012, for a review). For example, VanWormer (2004)
and Normand (2008) implemented treatment packages composed of pedometers, self-
monitoring, goal setting, and contingent praise to increase the number of daily steps taken.
Although the sample sizes were small, the reported interventions increased the number of
steps taken per day.
A recent study using procedures parallel to the contingency management reinforcement-
based procedures developed for reducing drug use (Peirce et al., 2006; Petry, Barry, Alessi,
Rounsaville, & Carroll, 2012; Petry, Martin, Cooney, & Kranzler, 2000; Petry, Weinstock,
& Alessi, 2011; Petry et al., 2005) found that interventions in which tangible reinforcers are
provided contingent on ambulatory activity can increase such activity. Petry, Andrade,
Barry, and Byrne (2013) randomized 45 sedentary older adults to either an intervention
comprising pedometers and guidelines to walk > 10,000 steps per day or to the same
intervention plus chances to win monetary prizes contingent upon meeting walking goals.
Participants randomized to the reinforcement contingency condition walked substantially
more, meeting target goals on 82.5% of days compared to 55.2% of days for those in the
control group. Furthermore, participants exposed to the reinforcement contingency showed
greater reductions in blood pressure and weight, as well as improvements in other fitness
indices, relative to participants in the non-reinforcement group.
Finkelstein, Brown, Brown, and Buchner (2008) also evaluated the efficacy of pedometers
and a monetary reinforcement procedure to increase walking in 51 adults. Participants
randomized to a treatment condition involving monetary reinforcement contingent upon
reaching walking goals were more active relative to participants in a control group, whose
behavior was not reinforced. However, this study lasted for only 4 weeks, so the question
remains as to whether the intervention would sustain walking levels over longer periods of
time. Furthermore, walking was reinforced only once, after study completion. Given that
delays to reinforcement reduce reinforcer effectiveness (e.g., Lussier, Heil, Mongeon,
Badger, & Higgins, 2006), reinforcing behavior more immediately might increase the
proportion of individuals who respond to a reinforcement intervention. In the Finkelstein et
al. (2008) study, only 38% of participants assigned to the reinforcement intervention met the
public health recommendations for moderate physical activity based on steps.
The monitoring schedule plays a prominent role in reinforcement-based interventions. In
substance abuse contingency management treatments based on these principles, monitoring
usually occurs at a relatively high frequency (e.g., twice or thrice weekly monitoring
schedules for 12 weeks; Lussier et al., 2006; Petry, 2000; Prendergast, Podus, Finney,
Greenwell, & Roll, 2006). Thus, opportunities for the behavior to be reinforced also occur
frequently. Such high-density reinforcement schedules exert strong control over behavior,
Andrade et al. Page 2
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
and they generate behavior change and maintain behavior while the contingency is in place.
These high-density schedules, however, are usually expensive and labor intensive. Further,
compared to less intensive schedules, high-density schedules appear very distinct from the
naturally occurring contingencies of reinforcement that may control the target behavior after
the intervention is withdrawn. Such discrepancies between experimenter controlled and
natural contingencies might reduce the likelihood of treatment generalization (Stokes &
Baer, 1977).
To increase the likelihood of generalization (i.e., maintenance of treatment effects), schedule
thinning, in which the density of reinforcement is gradually decreased over time, can be
incorporated after the target behavior is effectively modified (LeBlanc, Hagopian, Maglieri,
& Poling, 2002). One way to thin a reinforcement schedule is to implement more
intermittent schedules of reinforcement. A conceptually important feature of intermittent
schedules is the unpredictable availability of the reinforcer (Stokes & Baer, 1977). Use of an
intermittent schedule once a new behavior pattern has been established might promote
sustainable effects at relatively low costs, as the behavior is reinforced less frequently.
The present study evaluated the effects of reinforcement schedule thinning in the context of
a reinforcement intervention on the maintenance of increased ambulatory activity in
sedentary adults. The schedule thinning was comprised of a variable interval (VI) schedule
for which the interval gradually increased over time. In the context of this article, the term
“variable interval” is used to designate a monitoring system (and the corresponding
opportunity for reinforcement) that occurs. at intervals that were variable and increased over
time. Specifically, after achieving high rates of walking using a fixed-interval (FI)
monitoring plusreinforcement schedule, participants were randomly assigned to a
monitoring-only condition (with no tangible reinforcement contingent on exercising) or to a
monitoring plus reinforcement thinning condition. In this latter condition, the frequency with
which walking was monitored and reinforced decreased gradually over a 12-week period,
down to an average of once per month. The specific aim was to assess whether this
monitoring plus reinforcement thinning condition would sustain high rates of walking
relative to the monitoring-only condition throughout the period in which it was in effect. If
successful in maintaining behavioral gains, such a schedule would reduce the cost and time
burdens associated with frequent attendance required by fixed monitoring schedules. The
long-term effects of this monitoring plus reinforcement thinning system were also evaluated
to assess whether benefits on walking were maintained 9 weeks after the end of the
intervention period.
Method
Participants
Participants were recruited through advertisements stating that individuals were sought for a
study of methods to promote walking. Participants were eligible if they were 18 years or
older and walked fewer than 6,000 steps per day, on average, as assessed by a pedometer,
although this criterion was not disclosed to potential participants. This criterion is similar to
the index of sedentary activity used in another study (Petry et al., 2013) but slightly higher
than that applied in some other studies (<5,000, e.g., Tudor-Locke et al., 2008; Tudor-Locke
Andrade et al. Page 3
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
& Bassett, 2004). The 6,000-step criterion was used in this study to ensure the inclusion
criteria were not too stringent and to increase the potential for generalization of results to a
larger range of persons. Participants were ineligible if they had a major uncontrolled
psychiatric illness (e.g., psychosis, suicidality), had a physical condition that could interfere
with walking 10,000 steps per day (e.g., back or leg problem, recent heart attack), or were in
recovery from pathological gambling due to the potential similarity between gambling and
the treatment intervention (cf., Petry et al., 2006; Petry & Alessi, 2010). Initial screening
occurred over the phone, and potentially eligible individuals were scheduled for two in-
person assessment interviews, scheduled eight days apart. During the initial in-person
assessment, potential participants provided written informed consent, as approved by the
University Institutional Review Board.
Procedure
Baseline (week 0)—Those who were interested in the study and appeared eligible were
instructed at an initial baseline assessment to engage in their usual activities for the next 7
days while wearing a pedometer (Omron model HJ-112; Kyoto, Japan) at all times, except
when bathing and sleeping. This pedometer was chosen because it contains a memory
feature that records and stores total number of steps walked daily for up to 7 consecutive
days, and because it has been independently validated (Hasson, Haller, Pober,
Staudenmayer, & Freedson, 2009). This pedometer automatically resets step counts daily at
midnight, weighs 32 grams, and measures 7.3 cm long by 5.4 cm wide by 1.6 cm high.
After wearing the pedometer for 7 days, participants attended a second baseline assessment,
at which steps taken in the past week were evaluated. Those who walked >6,000 steps per
day on average were thanked for their time and provided additional resources regarding
methods to improve their physical activity levels. Those who walked <6,000 steps on
average completed the remainder of the structured baseline assessments and continued to the
3-week FI monitoring plusreinforcement phase of the study described below. Figure 1 shows
the flow of participants through the study phases. Participants were compensated with a $10
gift card for completing structured evaluations at baseline, week 3, week 15, and week 24,
with >93% of follow-ups completed (Figure 1). In these evaluations, information regarding
demographics, medical history, psychiatric distress, and physical activity levels was
collected.
Fixed interval (FI) monitoring plusreinforcement (weeks 1–3)—Following the
baseline assessment, all remaining eligible participants (n = 72) were exposed to a FI
monitoring-reinforcement condition for 3 weeks. They were instructed to continue wearing
the pedometer daily and encouraged to walk ≥10,000 steps per day. Participants were
scheduled to meet with a research assistant three times per week (e.g., Mondays,
Wednesdays and Fridays) for three consecutive weeks. In each 15-min meeting, pedometer
data were examined and reinforcement was delivered contingent on walking ≥10,000 steps
per day. (For the purposes of this study, we refer to this condition as FI monitoring
plusreinforcement phase because there are two intertwined schedules embedded in this
condition: the monitoring schedule and reinforcement schedule, both of which were fixed in
this phase. However, some might consider the reinforcement contingency to involve a
Andrade et al. Page 4
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
differential reinforcement of high rate behavior [DRH] schedule). The reinforcers were
opportunities to draw from a bowl and win prizes ranging from $1 to $100 in value. The
bowl contained 500 slips of papers, of which 50% were “winning” slips. Of these, 209 slips
(41.8%) were small prizes, 40 (8.0%) were large prizes, and 1 (0.2%) was a jumbo prize.
The other 250 (50.0%) non-winning slips were composed of an encouraging message,
“Good job!” Small prizes were worth about $1, such as food items, toiletries, and $1 gift
certificates. Large prizes were worth up to $20, and they consisted of retail items such as
clothing, watches, and gift cards to stores and restaurants. The jumbo prize was worth up to
$100, and consisted of items such as iPods, e-readers, and gift cards. Throughout the study,
new prizes were frequently made available according to participants’ preference.
All participants earned one draw for each day they walked ≥10,000 steps. To promote
sustained behavior change, participants also earned bonus draws if they walked ≥10,000
steps on the two to four consecutive days since their last visit. Bonus draws started at two,
and increased by two draws at each visit up to a maximum of eight draws. Bonus draws
were reset if patients failed to reach 10,000 steps on any day since the last visit or if patients
missed a scheduled appointment. Similar types of escalating schedules with a reset
contingency have been used effectively in contingency management treatment targeting
drug abstinence (e.g., Higgins, Wong, Badger, Haug Ogden, & Dantona, 2000; Petry et al.,
2005; Silverman, Robles, Mudric, Bigelow, & Stitzer, 2004).
At the end of the 3-week interval monitoring plus reinforcement phase, participants who
walked ≥10,000 steps per day on at least 14 of the 21 days were eligible to move to the
randomization phase. This subsample was chosen because these participants had
demonstrated initial behavior change and would therefore be in a position to demonstrate
durable behavior change. Those who failed to meet the 10,000 steps criterion on more than 7
of the 21 days were thanked for participation and informed about other methods to increase
walking (e.g., varying the routine, making it social, etc.), but they did not continue in the
study (see Figure 1).
Randomization (weeks 4–15)—Participants were randomized to a monitoring-only
condition or to a monitoring plus reinforcement thinning schedule for the next 12 weeks. To
ensure balance between the two conditions, a computerized urn randomization program
(Stout, Wirtz, Carbonari, & Del Boca, 1994) balanced group assignment based on whether
participants attended all sessions during the FI monitoring plus reinforcement phase and
whether they walked ≥10,000 steps on 18 or more of the 21 days during that phase.
Participants assigned to both conditions were instructed to continue wearing the pedometer
and were encouraged to walk ≥10,000 steps per day for the next 12 weeks. During this
phase, participants selected 2 potential meeting days each week separated by at least 72
hours (e.g., Mondays-Fridays, Mondays-Thursdays, or Tuesdays-Fridays) during which they
would be available to meet if the day was selected as a meeting day. Days were randomly
selected as meeting days, but participants were unaware of which days were randomly
selected as a meeting day until the morning of that day. In the mornings of randomly
selected meeting days, research staff contacted participants by phone and informed them that
they were due to meet that day.
Andrade et al. Page 5
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
Monitoring-only condition: Participants assigned to this condition earned a $5 gift card for
attending meetings on randomly selected meeting days. To earn the gift card, participants
also needed to bring their pedometers, with step data recorded for at least the past 4 days.
Receipt of this $5 gift card, however, was not contingent upon how many steps were
walked. Participants were congratulated for each day in which they walked ≥10,000 steps,
but tangible reinforcers were no longer provided.
Monitoring plus reinforcement thinning condition: Participants assigned to this condition
earned the same $5 gift card for attending randomly selected meeting dates and bringing the
pedometer with steps recorded in the past four days. In addition, these participants continued
earning “bonus” draws contingent on walking ≥10,000 steps on the prior four days. Bonus
draws increased for consecutive periods of time in which ≥10,000 steps were walked in the
past 4 days. The difference in this phase relative to the FI monitoring plus reinforcement
phase was that pedometer readings were not always available between contiguous meetings
(e.g., large gaps could occur between randomly selected meeting days; see below). Thus,
bonuses were earned so long as steps walked were ≥10,000 in at least the 4 days prior to the
randomly selected meeting date. Participants continued earning the same bonuses that they
had earned during the FI monitoring plusreinforcement phase (range, 2–8 draws). Bonuses
were reset if participants failed to attend a randomly selected meeting day or if steps
decreased below 10,000 on any of the 4 days prior to a randomly selected meeting date.
Once reset, bonuses could again escalate once walking resumed to ≥10,000 steps per day on
the 4 days prior to a randomly selected visit.
In both conditions, a Microsoft Excel macro determined specific meeting days for each
participant. The probabilities of scheduling a meeting on any potential day started at 50%
and decreased by 50% across every 4-week period, with the restriction that the total number
of scheduled visits was ≥7 during the 12-week period of the randomization phase (and
averaged 7.1 ± 0.6 for those assigned to the monitoring-only condition and 7.0 ± 0.3 for
those assigned to the monitoring plus reinforcement thinning condition). Specifically, the
probability of scheduling a meeting in any of the two potential meeting days during weeks
4–7, 8–11, and 12–15 was 50%, 25%, and 12.5%, respectively; on average, the number of
scheduled visits during these periods were four, two, and one. Participants were not
informed about the tapering schedule or average number of meetings; they were only told
that meeting days were randomly determined and could occur between 1 and 24 times over
the 12-week randomization phase.
Follow-up (weeks 16–24)—At the end of the intervention period, participants in both
conditions were given their pedometers to keep. They were encouraged to continue wearing
them to monitor their steps daily and to walk ≥10,000 steps per day. They were scheduled
for a 24-week follow-up evaluation, and a week before the evaluation, they were telephoned
and reminded of their upcoming appointment and to wear the pedometer daily for the week
preceding the evaluation.
Andrade et al. Page 6
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
Results
Initially, differences in baseline characteristics were evaluated between participants who
were later assigned to monitoring only and the monitoring plusreinforcement thinning
conditions. Independent t-tests were used for normally distributed continuous variables,
Mann Whitney U tests for non-normally distributed variables, and chi-square tests for
categorical variables. Two primary outcomes were defined a priori: (a) percentage of days
on which ≥10,000 steps were taken, and (b) average number of steps per day, each assessed
via pedometer readings. Initially, independent group t-tests evaluated differences in changes
in these walking indices between baseline and 3 weeks later, the period during which all
participants contacted reinforcement for walking ≥10,000 steps per day (i.e., the FI
monitoring plus reinforcement). Change scores were utilized (baseline minus post-baseline
values) because they were normally distributed.
The primary analyses focus on between-group differences in change from baseline scores
during the 12-week randomization phase. Again, independent group t-tests were used to
evaluate differences between the two treatment conditions. We also present descriptive data
on the average number of steps registered weekly throughout the study period. Missing data
throughout the randomization phase were not included, because they were relatively
infrequent and did not differ between treatment groups; only 14.8% and 9.2% of randomly
selected sessions were not attended in the monitoring-only and monitoring plus
reinforcement thinning condition, respectively, t (59) = 0.95, p = .35.
Finally, independent t-tests evaluated group differences in follow-up change from baseline
values for the primary walking data. These analyses were conducted twice, both considering
missing data as missing, and using a 0 (no change from baseline) for participants who failed
to complete the follow-up evaluation (n = 2 and 2 for each group). Because including
missing data as a 0 did not impact results, only analyses from follow-up completers are
presented. Thus, missing data were not included in any of the analyses herein reported.
Participant Characteristics
The sample comprised mostly females (90%), who described themselves as non-Hispanic
(95%) and White (82%). Mean (± standard deviation) age and annual income were 48
(±9.5) years and $59,913 (±$21,972), respectively. At baseline, participants walked on
average 4,444 (±1,108) steps per day. There were no significant differences between the
participants assigned to the monitoring only and monitoring plusreinforcement thinning
groups on any demographics or baseline characteristics.
Response to Experimental Contingencies
Figure 2 depicts the mean number of steps registered weekly across the 24-week study
period for each group. Figure 3 depicts data for individual participants in the monitoring
plus reinforcement thinning group and the monitoring-only group separately. The horizontal
lines indicate days on which > 10,000 steps were logged on the pedometer, and vertical
dashes indicate days on which monitoring visits were scheduled. Due to the memory
capacity of pedometers, there were missing data when monitoring visits were scheduled
Andrade et al. Page 7
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
more than seven days apart. Missing data occurred more often toward the end of the
randomization phase than in earlier parts of the phase because of the nature of the thinning
schedule, but missing data did not differ between groups, as noted earlier. As seen in Figure
3, participants from both groups exhibited similar patterns of walking ≥10,000 steps per day
during the BL and FI monitoring plus reinforcement phases (i.e., prior to randomization).
Only 4 participants had any days of walking ≥10,000 steps per day during BL, and in each
case, steps exceed 10,000 on only one day during BL. All participants walked substantially
more during the 3-week FI monitoring plusreinforcement phase, and they met walking goals
on almost all of the days of this phase. More specifically, the percent of days meeting target
goal was 96% and 92% for participants in the monitoring-only and monitoring plus
reinforcement thinning groups, respectively (see Table 1).
During the randomization phase, the overall performance from the participants in the two
groups differed. As indicated by the horizontal lines appearing in top graph relative to
bottom graph, participants exposed to the monitoring plus reinforcement thinning
contingency met walking goals more often than their counterparts exposed to monitoring
alone. In addition, participants in the monitoring plus reinforcement thinning group achieved
longer periods of walking > 10,000 steps per day than participants in the monitoring-only
group. For example, 19 of the 31 (61%) participants assigned to the monitoring plus
reinforcement thinning condition met walking goals for at least 3 consecutive weeks during
the randomization phase versus only 8 of 30 (26%) participants assigned to the monitoring-
alone condition.
Participants from both groups exhibited similar performance during the follow-up phase.
About half of the participants from each group (15 from the monitoring-only group and 16
from the monitoring plus reinforcement thinning group) met the ≥10,000 step goal on at
least one of the 7 days. Although days in which participants walked ≥10,000 steps during the
follow-up were lower than during the randomization phase, participants in both groups
registered more days with > 10,000 steps at follow-up than in baseline.
Visual inspection of the average group data (Figure 2) also shows that participants in the
monitoring plus reinforcement thinning group sustained higher levels of walking indices
throughout the randomization phase compared to those in the monitoring-only group.
However, activity levels in both groups seemed to decrease during the last 2–3 weeks of the
randomization phase, i.e., the period during which participants were exposed to the leanest
monitoring or monitoring plus reinforcement schedule (12.5% of chance of having a visit
scheduled). Table 1 depicts walking outcomes and statistical analyses comparing groups at
each phase of the study. There were no differences between groups during the pre-
randomization phase, when all participants contacted reinforcement for walking. Significant
differences between the two groups in changes from baseline emerged on both primary
walking indices during the randomization phase, ps < .004. The increases in the percent of
days participants walked ≥10,000 steps as well as the average number of steps walked per
day were significantly higher in the monitoring plus reinforcement thinning group compared
to the monitoring-only group. However, these significant between-group differences during
the randomization phase were not maintained at the 24-week follow-up evaluation.1
Andrade et al. Page 8
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
Reinforcement Earned and Adverse Events
During the 3-week FI monitoring plus reinforcement phase, participants who were later
assigned to the monitoring-only condition earned an average of 68 ± 13 draws, resulting in
$155 ± $48 in prizes, compared with an average of 64 ± 20 draws and $140 ± $59 in prizes
for those who were later assigned to the monitoring plus reinforcement thinning condition, t
(59) = 1.05 and 1.07, ps >.29. During the 12-week randomization phase, participants in the
monitoring plus reinforcement thinning condition earned an average of 36 ± 20 draws and
$77 ± $68 in prizes. No study-related adverse events occurred.
Discussion
This study found that continued reinforcement on a monitoring plus reinforcement thinning
schedule maintained ambulatory activity relative to an abrupt cessation of reinforcement
during the 12 weeks in which the thinning reinforcement schedule remained in effect.
Nevertheless, long-term walking outcomes were similar between conditions. Sustaining high
levels of walking beyond 12–15 weeks may require even longer durations of reinforcement-
based interventions.
Results from this study also demonstrate that walking, once established, can be maintained
with lower levels of reinforcement. On average, participants earned about $7 per day during
the initial reinforcement period (about $150 over 21 days). This amount was selected
because it is consistent with levels of reinforcement reported to alter substance use, weight
loss, and medication adherence (Petry et al. 2005; Petry, Barry, Pescatello, & White, 2011;
Petry, Rash, Byrne, Ashraf, & White, 2012), whereas lower monetary amounts appear
ineffective in engendering initial behavior change (Petry et al., 2004; Petry, Barry, et al.,
2012). Throughout the randomization phase of this study, less than $1 per day in
reinforcement ($77 over 84 days) was sufficient to sustain high rates of walking.
The monetary amount used during the randomization phase of this study was also lower than
that used in other randomized studies using monetary incentives to reinforce walking. In
Finkelstein et al.’s (2008) study, for example, participants could earn up to $150 during the
1In addition to visual inspection (Figures 2–3) and t-tests of between treatment groups effects (Table 1), multilevel modeling analyzed the step data as a discontinuous growth model. These analyses, conducted on SAS proc MIXED, used maximum likelihood estimation methods, in which missing data are taken into account in the process of estimating the covariance matrices (Singer & Willett, 2003). The time variable days (starting at baseline and running through the 24-week follow-up) was divided into study phases. Both the intercept and the day variables were included as random effects.The number of steps was influenced by the reinforcement offered during study phases. The effect from Baseline to the FI monitoring plus reinforcement phase was very large, reflecting the increase in steps taken when walking was reinforced, F (4536) = 253.11, p < .001. The slope of steps over time during this phase remained flat, F (4536) = 0.00, p >.90, indicating no change in steps during the 3-week FI monitoring plus reinforcement phase. As expected, no treatment condition effects emerged at this point, prior to randomization, in terms of differential level of steps, F (4536) = 1.97, p > .15, or slopes between groups, F (4536)=2.37, p > .15.The number of steps during the Randomization phase was also elevated with respect to baseline, F (4536) = 406.05, p < .001, and again the slope during this period was flat, F (4536) = 0.35, p >.50. At the transition to the Randomization phase, a significant treatment condition effect emerged, F (4536) = 4.22, p <.05, such that participants assigned to the monitoring plus reinforcement thinning condition evidenced a higher mean number of steps than those assigned to the monitoring-only condition. A significant treatment condition X slope effect also emerged during this period, F (4536) = 7.61, p < .01, accounted for by the decline over time in steps recorded by participants in the monitoring-only condition.Steps recorded at the 24-week follow-up were not significantly different from that recorded during the Randomization phase, F (4536) = 0.05, p > .80, and the slope during this 7-day period remained flat, F (4536) = 0.02, p > .80. There were no differences in steps, F (4536) = 0.01, p > .90 or treatment condition X slope effects during the follow-up period F (4536) = 0.01, p > .90. Data not reported; available from authors.
Andrade et al. Page 9
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
4 weeks of the study. In Petry et al.’s (2013) study, participants earned an average of $375
for increased walking during the 12-week intervention phase, and this study used a
monitoring-reinforcement procedure that resembled the one used in the initial reinforcement
phase of the current study. One important difference, however, was that participants were
not exposed to schedule thinning in the Petry et al. (2013) study; instead, they were
monitored and walking was reinforced on a set weekly schedule during the entire
intervention phase. The present study demonstrates that monitoring and reinforcement can
occur relatively infrequently yet sustain behavior change. During the randomization phase,
the average number of monitoring-reinforcement visits was only seven (four, two, and one
in each 4-week period), and participants earned a total of $77 dollars in prizes during this
phase. The density of reinforcement during each consecutive four-week period was on
average $44, $22, and $11, respectively.
Data from the FI monitoring plus reinforcement phase of this study show that the
reinforcement contingencies substantially increased the number of steps taken per day.
Compared to baseline, participants, on average, increased their steps per day by about 6,000
steps, up from 4,000 per day at baseline to 10,000 per day during this phase. These results
suggest that a program comprising pedometers, daily step goals, monitoring, and tangible
reinforcers can promote increased levels of walking. Eighty-five percent (61 of 72) of
individuals exposed to the FI monitoring plus reinforcement procedures walked ≥10,000
steps on at least two-thirds of the days and thus qualified for the randomization phase of the
study.
Using a randomized design, Finkelstein et al. (2008) also reported increased activity levels
among participants who received reinforcement for reaching walking goals compared to
participants who did not. In that study, however, the percentage of participants who met the
public health guidelines for moderate physical activity was substantially lower than in the
current study (38% vs. 85%). This difference could relate to many factors including different
populations and settings, or to design characteristics of the reinforcement interventions. For
example, Finkelstein et al. provided slightly lower magnitude reinforcement than the initial
FI monitoring plus reinforcement phase in the present study. Further, in the Finkelstein et al.
study, all reinforcement was provided at the end of the study period, whereas reinforcement
occurred up to three times per week in the FI monitoring plus reinforcement phase of the
current study.
The studies by VanWormer (2004) and Normand (2008) demonstrated that a self-
management treatment package could increase the number of steps taken by participants,
even without programmed reinforcement contingencies. Direct comparisons between these
studies and the current one, however, are difficult due to methodological differences. For
example, in the current study the target behavior (i.e., walking goals) was the same for all
participants (≥10,000 steps), whereas in the other studies the walking goals varied widely
across participants in an individualized manner. Further, the prior studies were of shorter
durations than this evaluation.
In the current study, participants increased walking by approximately 6,000 steps per day
under the reinforcement contingencies. Whether this large increase in physical activity
Andrade et al. Page 10
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
produces health benefits that outweigh its costs is an empirical question, but the present
study demonstrates methods that can minimize personnel and reinforcement costs while
maintaining behavior change. Although future studies are needed to identify the most
efficacious and least time intensive approaches to delivering reinforcement, this study is
among the first to address the minimum frequency of monitoring and reinforcement
necessary for maintaining clinically important behavior change. Studies that implement
similar procedures, such as contingency management treatments for substance use disorders,
might implement schedule-thinning procedures to maintain treatment gains as well.
The ecological validity of the current study might also be questioned. Although the sample
resembles those in other observational studies and randomized clinical trials (see Bravata et
al. 2007), at least regarding gender and age, we cannot determine whether the effects
observed here will generalize to other populations or settings. Some other limitations should
be considered when interpreting the results from this study. First, the sample was composed
primarily of white well-educated women with middle incomes or higher, so the findings
might not generalize to men or other less educated groups or to individuals of other racial or
ethnic groups. Furthermore, participants responded voluntarily to advertisements and were
willing to be available on two to three potential meeting days each week. Results might
differ with individuals who do not self-select to participate in programs that enhance
walking. Nevertheless, the average number of steps taken daily at baseline by this sample
was lower than U.S. national average and below the average number of steps taken by
women of the same age range in general (Bassett, Wyatt, Thompson, Peters, & Hill, 2010).
Other limitations should be considered in addition to those related to the sample
characteristics. The efficacy of reinforcement to initiate behavior change was not directly
assessed in this study, and subsequent studies should evaluate the minimal reinforcement
levels needed to engender walking at high rates. Furthermore, objective physical
measurements (e.g., weight, blood pressure) were not taken, and thus this study did not
determine if the observed increases in walking impacted physical health or fitness indices. It
is also possible that participants may have given pedometers to others to wear, although
none reported doing so at the follow-up evaluation. Including individuals who self selected
to increase walking and conducting assessments in-person (as opposed to remote
computerized uploading of pedometer readings) may guard against “cheating” in this
context, but the possibility of deceit must always be considered when designing and
implementing reinforcement interventions (Petry, 2012).
Although this research relied on government funding, the current procedures may eventually
facilitate adoption of this type of intervention. Because the cost was low, some individuals
interested in increasing and sustaining high levels of exercise may be willing to fund their
own treatment. For example, participants could make monetary deposits that would be
reimbursed contingent upon meeting the target goals. Alternatively, employers, health care
or other organizations (e.g., retirement communities or schools) may cover costs of
reinforcers if improved performance, health, or other outcomes were noted in conjunction
with increased activity levels.
Andrade et al. Page 11
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
In summary, this study demonstrates that a monitoring plus reinforcement thinning schedule
using the prize reinforcement system has the potential to maintain high rates of ambulatory
activity in sedentary adults. Effects were achieved with relatively low levels of
reinforcement, delivered at infrequent intervals. These aspects of the intervention are likely
to enhance the dissemination and acceptability of contingency-management interventions
more generally, and these interventions might ultimately prove to be cost-beneficial for
improving health, especially in high-risk patient populations.
Acknowledgments
We thank Amy Novotny for assistance in conducting this study. This research and preparation of this report were funded in part by NIH grants P30-DA023918, R01-DA027615, R01-DA022739, R01-DA13444, P50-DA09241, P60-AA03510, R01-HD075630, R01-DK097705, and T32-AA07290.
References
Bassett DR, Wyatt HR, Thompson H, Peters JC, Hill JO. Pedometer-measured physical activity and health behaviors in U.S. adults. Medicine and Science in Sports and Exercise. 2010; 42:1819–1825. [PubMed: 20305579]
Boone-Heinonen J, Evenson KR, Taber DR, Gordon-Larsen P. Walking for prevention of cardiovascular disease in men and women: a systematic review of observational studies. Obesity Reviews. 2009; 10:204–217. [PubMed: 19207874]
Bravata DM, Smith-Spangler C, Sundaram V, Gienger AL, Lin N, Lewis R, Sirard JR. Using pedometers to increase physical activity and improve health: A systematic review. The Journal of the American Medical Association. 2007; 298:2296–2304. [PubMed: 18029834]
Finkelstein EA, Brown DS, Brown DR, Buchner DM. A randomized study of financial incentives to increase physical activity among sedentary older adults. Preventive Medicine. 2008; 47:182–187. [PubMed: 18571226]
Garber CE, Blissmer B, Deschenes MR, Franklin BA, Lamonte MJ, Lee I, Swain DP. Quantity and quality of exercise for developing and maintaining cardiorespiratory, musculoskeletal, and neuromotor fitness in apparently healthy adults: Guidance for prescribing exercise. Medicine and Science in Sports and Exercise. 2011; 43:1334–1359. [PubMed: 21694556]
Haskell WL, Lee I, Pate RR, Powell KE, Blair SN, Franklin BA, Bauman A. Physical activity and public health: Updated recommendation for adults from the American College of Sports Medicine and the American Heart Association. Medicine and Science in Sports and Exercise. 2007; 39:1423–1434. [PubMed: 17762377]
Hasson RE, Haller J, Pober DM, Staudenmayer J, Freedson PS. Validity of the Omron HJ-112 pedometer during treadmill walking. Medicine and Science in Sports and Exercise. 2009; 41:805–809. [PubMed: 19276853]
Higgins ST, Wong CJ, Badger GJ, Haug Ogden DE, Dantona RL. Contingent reinforcement increases cocaine abstinence during outpatient treatment and 1 year of follow-up. Journal of Consulting and Clinical Psychology. 2000; 68:64–72. [PubMed: 10710841]
Hu FB, Li TY, Colditz GA, Willett WC, Manson JE. Television watching and other sedentary behaviors in relation to risk of obesity and type 2 diabetes mellitus in women. The Journal of the American Medical Association. 2003; 289:1785–1791. [PubMed: 12684356]
Hu FB, Sigal RJ, Rich-Edwards JW, Colditz GA, Solomon CG, Willett WC, Manson JE. Walking compared with vigorous physical activity and risk of type 2 diabetes in women: a prospective study. The Journal of the American Medical Association. 1999; 282:1433–1439. [PubMed: 10535433]
LeBlanc LA, Hagopian LP, Maglieri KA, Poling A. Decreasing the intensity of reinforcement-based interventions for reducing behavior: Conceptual issues and a proposed model for clinical practice. The Behavior Analyst Today. 2002; 3:289–300.
Andrade et al. Page 12
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
Le-Masurier GC, Sidman CL, Corbin CB. Accumulating 10,000 steps: Does this meet current physical activity guidelines? Research Quarterly for Exercise and Sport. 2003; 74:389–394. [PubMed: 14768840]
Lussier JP, Heil SH, Mongeon JA, Badger GJ, Higgins ST. A meta-analysis of voucher-based reinforcement therapy for substance use disorders. Addiction. 2006; 101:192–203. [PubMed: 16445548]
Normand MP. Increasing physical activity through self-monitoring, goal setting, and feedback. Behavioral Interventions. 2008; 23:227–236.
Peirce JM, Petry NM, Stitzer ML, Blaine J, Kellogg S, Satterfield F, Li R. Effects of lower-cost incentives on stimulant abstinence in methadone maintenance treatment: A National Drug Abuse Treatment Clinical Trials Network study. Archives of General Psychiatry. 2006; 63:201–208. [PubMed: 16461864]
Petry NM. A comprehensive guide to the application of contingency management procedures in clinical settings. Drug and Alcohol Dependence. 2000; 58:9–25. [PubMed: 10669051]
Petry, NM. Contingency Management for Substance Abuse Treatment: A Guide to Implementing this Evidence-based Practice. New York, NY: Routledge/Taylor & Francis; 2012.
Petry NM, Alessi SM. Prize-based contingency management is efficacious in cocaine-abusing patients with and without recent gambling participation. Journal of Substance Abuse Treatment. 2010; 39:282–288. [PubMed: 20667679]
Petry NM, Andrade LF, Barry D, Byrne S. A randomized study of reinforcing walking in older adults. Psychology and Aging. 2013; 28:1164–1173. [PubMed: 24128075]
Petry NM, Barry D, Alessi SM, Rounsaville BJ, Carroll KM. A randomized trial adapting contingency management targets based on initial abstinence status of cocaine-dependent patients. Journal of Consulting and Clinical Psychology. 2012; 80:276–285. [PubMed: 22229758]
Petry NM, Barry D, Pescatello L, White WB. A low-cost reinforcement procedure improves short-term weight loss outcomes. The American Journal of Medicine. 2011; 124:1082–1085. [PubMed: 21851917]
Petry NM, Kolodner KB, Li R, Peirce JM, Roll JR, Stitzer ML, Hamilton JA. Prize-based contingency management does not increase gambling. Drug and Alcohol Dependence. 2006; 83:269–273. [PubMed: 16377101]
Petry NM, Martin B, Cooney JL, Kranzler HR. Give them prizes, and they will come: Contingency management for treatment of alcohol dependence. Journal of Consulting and Clinical Psychology. 2000; 68:250–257. [PubMed: 10780125]
Petry NM, Peirce JM, Stitzer ML, Blaine J, Roll JM, Cohen A, Li R. Effect of Prize-Based Incentives on Outcomes in Stimulant Abusers in Outpatient Psychosocial Treatment Programs: A National Drug Abuse Treatment Clinical Trials Network Study. Archives of General Psychiatry. 2005; 62:1148–1156. [PubMed: 16203960]
Petry NM, Rash CJ, Byrne S, Ashraf S, White WB. Financial reinforcers for improving medication adherence: Findings from a meta-analysis. The American Journal of Medicine. 2012; 125:888–896. [PubMed: 22800876]
Petry NM, Tedford J, Austin M, Nich C, Carroll KM, Rounsaville BJ. Prize reinforcement contingency management for treating cocaine users: How low can we go, and with whom? Addiction. 2004; 99:349–360. [PubMed: 14982548]
Petry NM, Weinstock J, Alessi SM. A randomized trial of contingency management delivered in the context of group counseling. Journal of Consulting and Clinical Psychology. 2011; 79:686–696. [PubMed: 21806297]
Prendergast M, Podus D, Finney J, Greenwell L, Roll J. Contingency management for treatment of substance use disorders: A meta-analysis. Addiction. 2006; 101:1546–1560. [PubMed: 17034434]
Silverman K, Robles E, Mudric T, Bigelow GE, Stitzer ML. A randomized trial of long-term reinforcement of cocaine abstinence in methadone-maintained patients who inject drugs. Journal of Consulting and Clinical Psychology. 2004; 72:839–854. [PubMed: 15482042]
Singer, JD.; Willett, JB. Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence. London: Oxford University Press; 2003.
Andrade et al. Page 13
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
Stokes TF, Baer DM. An implicit technology of generalization. Journal of Applied Behavior Analysis. 1977; 10:349–367. [PubMed: 16795561]
Stout RL, Wirtz PW, Carbonari JP, Del Boca FK. Ensuring balanced distribution of prognostic factors in treatment outcome research. Journal of Studies on Alcohol and Drugs. 1994; 12:70–75.
Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, McDowell M. Physical activity in the United States measured by accelerometer. Medicine and Science in Sports and Exercise. 2008; 40:181–188. [PubMed: 18091006]
Tudor-Locke C, Bassett DR. How many steps/day are enough? Preliminary pedometer indices for public health. Sports Medicine. 2004; 34:1–8. [PubMed: 14715035]
Tudor-Locke C, Craig CL, Brown WJ, Clemes SA, De Cocker K, Giles-Corti B, Blair SN. How many steps/day are enough? For adults. International Journal of Behavioral Nutrition and Physical Activity. 2011; 8:79. Retrieved from: http://www.ijbnpa.org/content/8/1/79. [PubMed: 21798015]
Tudor-Locke C, Hatano Y, Pangrazi RP, Kang M. Revisiting “How many steps are enough?”. Medicine and Science in Sports and Exercise. 2008; 40:S537–S543. [PubMed: 18562971]
Van Camp CM, Hayes LB. Assessing and increasing physical activity. Journal of Applied Behavior Analysis. 2012; 45:871–875. [PubMed: 23322945]
VanWormer JJ. Pedometers and brief e-counseling: Increasing physical activity for overweight adults. Journal of Applied Behavior Analysis. 2004; 37:421–425. [PubMed: 15529901]
World Health Organization. Global Recommendations on Physical Activity for Health. Geneva, Switzerland: WHO; 2010. http://www.who.int/dietphysicalactivity/factsheet_recommendations/en/
Andrade et al. Page 14
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
Figure 1. Flowchart of participants in the study.
Andrade et al. Page 15
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
Figure 2. Average number of steps registered weekly across the study. Filled symbols refer to
participants randomized to the monitoring plus reinforcement thinning condition during the
Randomization phase, and unfilled symbols refer to participants randomized to the
monitoring-only condition during the Randomization phase; all participants received
reinforcement during the FI monitoring plus reinforcement phase. Values represent group
means collected each week, but not all participants provided data at each week during the
Randomization phase. During the Randomization phase, participants met with research staff
on average 4 visits during the first four week period (weeks 4–7), on average twice during
the second four week period (weeks 8–11), and on average once during the last four week
period (weeks 12–15). See text for further details. BL = Baseline
Andrade et al. Page 16
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
Figure 3. Consecutive days each participant in the monitoring-only and monitoring plus reinforcement
thinning groups met walking goals. Horizontal lines depict days on which > 10,000 steps
were logged into the pedometer of each participant across conditions. Vertical dashes
represent days monitoring visits were scheduled. The thinner horizontal lines depict missing
data that were considered as meeting the walking criteria when the monitoring visits were
scheduled more than 7 days apart and if the participant met walking goals on both ends of
the data string. In each panel, participants are arranged with those showing the greatest
Andrade et al. Page 17
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
number of days meeting walking goals on the top to participants showing the least number
of days meeting walking goals on the bottom. The numerals on the y-axis represent
participant numbers. See text for further details.
Andrade et al. Page 18
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
Author M
anuscriptA
uthor Manuscript
Andrade et al. Page 19
Tab
le 1
Prim
ary
wal
king
out
com
es o
btai
ned
at e
ach
asse
ssm
ent.
Var
iabl
e
FI
mon
itor
ing
rein
forc
emen
t(W
ks 1
–3)
Stat
isti
cs (
df)
Ran
dom
izat
ion
(Wks
4–1
5)St
atis
tics
(df
)F
ollo
w-u
p(W
eek
24)
Stat
isti
cs (
df)
Perc
ent d
ays
wal
ked
≥10,
000
step
st (
59)=
1.7
3, p
=.0
9t (
59)
= 3
.88,
p<
.001
t (55
) =
0.67
, p=
.51
Mon
itori
ng-o
nly
96.1
(6.
0)55
.3 (
31.0
)31
.1 (
37.5
)
M +
R th
inni
ng91
.6 (
11.4
)82
.6 (
23.5
)24
.9 (
33.0
)
Mea
n st
eps/
day
t (59
) =
0.1
6, p
=.8
8t (
59)
= 2
.98,
p=
.004
t (55
) =
.29,
p=
.77
Mon
itori
ng-o
nly
10,5
71 (
721)
8,42
8 (2
,033
)6,
754
(3,1
59)
M +
R th
inni
ng10
,349
(68
2)9,
561
(1,5
70)
6,49
8 (2
,408
)
FI=
Fix
ed in
terv
al. M
+ R
thin
ning
= M
onito
ring
plu
s re
info
rcem
ent t
hinn
ing.
Val
ues
repr
esen
t mea
ns (
stan
dard
dev
iatio
ns).
Sta
tistic
s re
fer
to d
iffe
renc
es b
etw
een
trea
tmen
t gro
ups
with
res
pect
to c
hang
e fr
om b
asel
ine
valu
es.
J Appl Behav Anal. Author manuscript; available in PMC 2016 March 04.
