Objective. Elevated television (TV) viewing and physical inactivity promote obesity in children. Thus, changes in physical activity and sedentary behavior seem critical to treating childhood obesity.
Present Study. Using a randomized, 2-arm design, this pilot study tested the effects of contingent TV on physical activity and TV viewing in 10 obese children. TV viewing was contingent on pedaling a stationary cycle ergometer for experimental participants but was not contingent on pedaling for control participants. The study was conducted over 12 weeks, including a 2-week baseline period.
Results. Multivariate analyses indicated that the intervention significantly increased pedaling and reduced TV-viewing time. During the treatment phase, the experimental group pedaled 64.4 minutes per week on average, compared with 8.3 minutes by controls. The experimental group watched 1.6 hours of TV per week on average, compared with 21.0 hours per week on average by controls during this phase. Secondary analyses indicated that the experimental group showed significantly greater reductions in total body fat and percent leg fat. Total pedaling time during intervention correlated with greater reductions in percent body fat (r = −0.68).
Conclusions. Contingencies in the home environment can be arranged to modify physical activity and TV viewing and may have a role in treating childhood obesity. Contingent TV may be one method to help achieve this goal.
The prevalence of childhood obesity has increased substantially in recent decades,1,2 a concerning finding given its associated health complications.3,4 It is known that obesity is environmentally influenced in part5,6 and that a sedentary lifestyle and sustained physical inactivity may be risk factors for obesity in youth.7 Of particular concern is television (TV) viewing, which has been associated with childhood obesity in both cross-sectional8 and longitudinal studies.9,10 These data suggest the potential utility of targeting TV-viewing practices for treating childhood obesity.
Behavioral interventions have been generally effective in producing short- and long-term weight loss in obese children.11–13Although data are limited, recent behavioral programs have been developed that specifically target reductions in TV-viewing time.14,15 These approaches rely on methods such as information dissemination and restricted TV accessibility. An alternative approach has attempted to capitalize on the reinforcing value of TV to increase children's physical activity.16,17 Based on behavioral theory,18Epstein and colleagues demonstrated that children would increase pedaling on a cycle ergometer to obtain access to contingent TV. For example, by making movie viewing and video game playing contingent on pedaling a cycle ergometer, Saelens and Epstein17successfully increased obese children's biking in the laboratory.
The present pilot study tested the effects of a contingent TV system on obese children's physical activity and TV-viewing time in the home environment over several months. The intervention consisted of a cycle ergometer that electrically interfaced with a TV so that pedaling was necessary to activate the TV. This contingent TV system, colloquially called the “TV cycle,” used TV as a continuous and immediate incentive to increase children's physical activity in the home. It was hypothesized that, compared with a control condition in which TV accessibility was not contingent on pedaling, the experimental group's physical activity would increase and TV viewing would decrease over a 12-week period. Secondary analyses compared changes in body fat between the 2 groups.
Ten obese children 8 to 12 years old participated in the study. The sample included 3 girls and 7 boys. Eligible participants: 1) had a body mass index above the 85th percentile for age and sex from the first National Health and Nutrition Examination Survey19; 2) watched at least 2 hours of television per day as assessed by self-report; 3) did not engage in regular physical activity, as indicated by parental report; 4) went home directly after school most days of the week, so that the device could achieve a measurable effect; 5) had no siblings within 7 years of age, so as to avoid sibling use of the TV cycle or sibling coercion; and 6) had sufficient physical space within the home to install the TV cycle.
Parents of children meeting these initial criteria were sent a second-screen questionnaire, including: 1) the Brief Symptom Inventory20 to screen for parental psychopathology, and 2) the Child Behavior Checklist21 to screen for child internalizing and externalizing behavioral disorders. Participant characteristics are summarized in Table 1.
Advertisements were placed in local newspapers offering an experimental treatment program for overweight children who watch excessive TV. Interested parents called the Obesity Research Center and were administered the preliminary phone screen (previously described). Families passing this screen were mailed the second-screen questionnaire packet for final determination of eligibility. Eligible participants and their parents came to the laboratory for study orientation, cycle ergometery fitness testing (described below), and baseline body composition measurement (described below). Participants were randomly assigned to either the control or treatment condition using a computer pseudo-random number generator. The contingent TV was placed in the child's home within 1 week. To maximize protocol compliance and minimize inconveniences to other family members, parent-controlled locks were placed on other household TVs. Although parents conceivably could allow the participant to watch their own TV, this was strongly discouraged by the investigators. For this pilot study, we did not quantify the extent to which participants might have watched other TVs in the home. Based on informal weekly interviews with parents, this was not a substantial problem and was a very rare occurrence.
The study length was 12 weeks, including a 2-week baseline phase and a 10-week treatment phase. For all participants, the TV cycle was set to the control condition for the initial 2-week baseline condition. In this mode, the contingency between pedaling and TV viewing was not activated. For the control group, the contingency remained inactive throughout the entire intervention. For the experimental group, the contingency was activated after the 2-week baseline phase (participants were told that this would occur on their enrollment in the study and assigned to this condition). In this condition, TV viewing was contingent on pedaling at the prescribed intensity level or greater. Otherwise, the system was left in the home for the child to use as much as he/she chose. No other instructions were provided. After completing participation, the TV cycle was removed from the house and posttest body composition measures were taken. Participants were given a $10 gift for every month of participation.
Design of the Contingent TV
The contingent TV consisted of 3 interfaced components: a standard TV, a custom designed power controller, and a LifeCycle (LifeFitness, Franklin Park, IL) 3500 electronically braked cycle ergometer. The power controller measured revolutions per minute (RPM) of cycling and activated power to the TV contingent on the child pedaling at or above the prescribed work level. It consisted of a printed circuit board, a small liquid crystal display, and several push buttons for user input. The power controller was housed along with a 12-volt power supply in a locking case, which also contained the TV power cord. Power control was maintained with a printed circuit board completed with 8-bit microcontroller, real-time clock, and power-switching circuitry. The microcontroller software recorded pedaling rate and, when participants pedaled at a rate corresponding to at least 50% of their maximal oxygen consumption (V˙o2max), signaled the power switching circuitry to activate the TV.
For participants assigned to the experimental condition, the microcomputer was set to a connect mode that activated the contingency between TV viewing and pedaling at the prescribed work level. For the first 3 participants in the experimental condition, the device was set at a reinforcement ratio of 1:1, such that 1 minute of pedaling earned 1 minute of TV viewing. Because the first 3 participants perceived the 1:1 ratio as requiring too much effort to sustain, we increased the reinforcement ratio to 1:2 for the remaining experimental participants (ie, 1 minute of pedaling earned 2 minutes of TV viewing). There was no significant difference between the 2 ratio conditions with respect to outcome measures. For the control group, the contingency was not activated. The microcomputer continuously recorded the following information during each pedaling bout: date, time of pedaling initiation, average RPM achieved, maximum RPM achieved, and duration of pedaling. These data were uploaded at weekly home visits.
Fitness Testing for Exercise Prescription
The fitness of each participant was assessed using aV˙o2max bicycle ergometer test,22 an accepted index of aerobic fitness children.23 An electronically braked ergometer (Lifecycle 3500) interfaced with a control panel maintained a specified level of resistance as the child pedaled. Each level of intensity lasted 2 minutes with the cadence rate set between 60 and 70 RPM. Children began pedaling at a work rate of 25 W (level 1), after which the work rate was increased to 60 W (level 2), 80 W (level 3), and successive 20-W increments until volitional exhaustion was reached. Testing was conducted with a multichannel 12 lead electrocardiograph (Quinton Q3000 Monitor, Quinton Instruments, Bothell, WA). Respiratory gases (oxygen consumption and carbon dioxide production) were collected using a Hans Rudolph 2-way face mask (Sensormedics, Yorba Linda, CA), interfaced with a metabolic cart containing Beckman infrared analyzers.V˙o2max levels were monitored for a plateau (or peak V˙o2) during fitness testing as further indication of volutional exhaustion.
The level on the bicycle corresponded to 50% ofV˙o2max as the minimum rate at which participants needed to pedal to activate the contingency. This workload generally corresponds to a moderate intensity level, which confers desirable health benefits.24 A more intense workload was not prescribed because of concerns that obese, sedentary children would not comply with such a regimen. Moreover, there is limited information on the exercise intensity level(s) or schedules of incremental intensity that might be optimal for treating obese children.25 In the present sample, 50% of theV˙o2max corresponded to ∼2.7 metabolic equivalent units.26
The primary outcome variables were pedaling and TV-viewing times, which were continuously recorded by the micro-computer of the TV cycle. Outcomes variables for secondary analyses were body mass index (kg/m2) and percent total body fat measured by either dual-energy x-ray absorptiometry (DXA)27 or bioimpediance analysis (Body Composition Analyzer [BCA] model 680, Schaumburg, IL),28 using the equation of Kushner et al.29 DXA was performed on a Lunar DPX-L scanner 1.5E (Lunar Corp, Madison, WI) equipped with Pediatric Software, Version 1.5e (Lunar Corp, Madison, WI). DXA also yielded measures of arm and leg percent fat. One control participant was unable to complete the follow-up DXA measurement but had total body fat measured by bioimpediance analysis. One experimental participant did not return to the laboratory after treatment for logistic reasons and, hence, follow-up measurements were not obtained. In all, changes in total body fat were measured on 9 children (ie, 5 experimental and 4 control participants); changes in arm and leg fat were measured on 8 participants (ie, 5 experimental and 3 control participants).
Descriptive data on TV-viewing and pedaling times were computed for weeks 1 to 2, 3 to 5, 6 to 8, and 9 to 12 (to illustrate gradual changes over time). A 2 (condition) × 4 (time) multivariate analysis of variance (MANOVA) on pedaling and TV-viewing time was conducted. Based on a significant MANOVA, 2 separate 2 (condition) × 4 (time) analyses of variance tested the effect of the intervention on pedaling and TV-viewing times, respectively, across the 4 time blocks. Significant Condition × time interactions were probed with posthoc comparisons by constructing and testing contrasts of interest. For secondary analyses, t tests compared the groups on changes in body composition from baseline to posttest. Pearson's correlations between total pedaling time and changes in body fat also were conducted. All significance tests were 2-sided with α set at 0.05.
Data on pedaling and TV-viewing time during each of the 4 time blocks for each group is presented in Table 2. Not all participants completed all 12 weeks of the study. The modal number of weeks completed by participants was 12; however, 2 participants withdrew before the eighth week (1 family terminated early because of a mechanical problem with the Lifecycle, the other because of unanticipated family travel that forced them to withdraw early). The MANOVA testing the intervention's joint effect on pedaling and TV-viewing times indicated significant effects for condition (Hotelling's T = 0.95; P< .0001), time (Hotelling's T = 0.76;P < .0001), and the condition × time interaction (Hotelling's T = 0.50; P < .0001). The interaction is depicted in Fig 1, which illustrates the changes in pedaling and TV-viewing times across the 4 time blocks.
The univariate analysis of variance of pedaling time indicated a significant effect for condition (F[1,92] = 6.53;P = .01) with the treatment and control groups, respectively, pedaling 50.5 and 17.7 minutes per week on average across the entire 12-week study. The condition × time interaction approached significance (F[3,92] = 2.31; P= .08). Posthoc analyses indicated that, among the experimental group, pedaling time significantly increased from baseline to weeks 3 to 5 (t = 1.99; P = .05). Changes in pedaling from baseline to weeks 6 to 8 (t = 0.21;P = .83) and 9 to 12 (t = 1.18;P = .24) were not significant. Among the control group, changes in pedaling from baseline to weeks 3 to 5 (t = −1.42; P = .16) and 6 to 8 (t = −1.82; P = .09) were not significant. Changes from baseline to weeks 9 to 12 were significant (t = −2.35;P = .05). Across the 10-week treatment phase (ie, weeks 3–12), the experimental and control groups pedaled an average of 64.4 and 8.3 minutes, respectively (t = 2.6;P = .04).
The univariate analysis of variance of TV-viewing time indicated significant effects for condition (F[1,94] =76.11;P < .0001), time (F[3,94] = 8.06;P < .0001), and the condition × time interaction (F[3,94] = 10.73; P < .0001). The treatment and control groups watched an average of 5.4 and 21.3 hours per week of TV across the entire 12-week study. Posthoc analyses indicated that, among the treatment group, TV-viewing time significantly decreased from baseline to weeks 3 to 5 (t = −6.97; P < .0001), weeks 6 to 8 (t = −7.08; P < .0001), and weeks 9 to 12 (t = −7.14; P < .0001). Among the control group, changes in TV viewing from baseline to weeks 3 to 5 (t = 1.08; P = .28), weeks 6 to 8 (t = 0.75; P = .46), and weeks 9 to 12 (t = 0.45; P = .65) were not significant. Across the 10-week treatment phase (ie, weeks 3–12), the experimental and control groups watched an average of 1.6 and 21.0 hours of TV per day, respectively (t = −6.42;P = .006).
Secondary analyses indicated that, compared with the control group, the experimental group showed significantly greater reductions in percentage leg fat (t = −2.46; P = .05) and percentage total body fat (t = −2.24;P = .06). The experimental group reduced total percent body fat (−1.2% on average), whereas the control group increased percent body fat (0.9% on average). Similarly, the experimental group reduced percent leg fat (−1.6% on average), whereas the control group increased percent leg fat (0.7% on average). Across all participants, there was a significant negative association between total pedaling time during the entire study and change in percent body fat (Pearson'sr = −0.68; P = .04). Similar results emerged when correlating pedaling time during weeks 3 to 12 (ie, after baseline) with change in percent body fat (r = −0.73;P = .03).
Introduction of the contingent TV significantly increased physical activity and reduced TV viewing among participants. During the treatment phase, the experimental group pedaled 64.4 minutes per week on average compared with 8.3 minutes on average pedaled by the control group. Also, during this phase, the experimental group watched 1.6 hours of TV per week on average, compared with the 21.0 hours on average watched by the control group. To put these numbers in perspective, the general population of 2- to 11-year-old children watches ∼23 hours of TV per week.30 Thus, the intervention had a significant effect on TV-viewing time. The lack of pedaling by the control group is consistent with behavioral theory, given that there was no sustained incentive for such behavior. This finding is consistent with anecdotal observations that placing a stationary cycle in front of a TV, without sustained incentives to pedal, will only have a transient influence on behavior.
These encouraging results illustrate the potential benefits of adopting behavioral theory to guide clinical interventions targeting physical activity and TV viewing.31–33 The contingent TV system used established behavioral principles to alter the association of TV viewing with inactivity in the home. The fact that significant behavior change was observed in a small sample treated for a relatively short time demonstrates the potential power of environmental modification for inducing immediate change. Although measures of hedonic response to exercise were not obtained on participants, anecdotally the intervention was well accepted by parents and children. None of the families reported dissatisfaction with the treatment, including the 2 families that withdrew early.
Several caveats concerning this study deserve mention. First, we did not measure how children reallocated their residual TV-viewing time, that is, what they did instead of watching TV. Larger studies might explore this important issue through physical activity recalls, posttreatment fitness testing, and/or more thorough body composition analyses. Children might substitute TV viewing with other sedentary behaviors (ie, substitute reinforcers),34,35 or they might go to friend's homes to watch TV. In contrast, the increased cycling may generalize to other physical activities in or out of the home. Laboratory experiments suggest that children who are reinforced for reduced sedentary behavior do not simply substitute one sedentary behavior for another but reallocate a certain portion of time to other physical activities.36 These questions address the important issue of alternative choices for activity/inactivity and can be readily studied in future studies from a behavioral economic perspective.37
Second, the observed body composition effects should be interpreted cautiously. Given the total time experimental participants spent pedaling on average, these children arguably reduced caloric intake and/or increased other physical activities during the course of the study to achieve body composition changes. We did not measure other behavior changes, which is critical for future studies. Particularly relevant may be changes in energy intake that might have cooccurred with intervention because TV viewing may promote snacking and caloric intake in children.9,38 A recent study found that, among children assigned to a family-based behavioral intervention that targeted parental involvement, greater reductions in TV room, living room, or bedroom eating were associated with greater weight loss.39 To the extent that the present intervention reduced TV viewing, it may have also reduced contextual cues for eating.
A final limitation is that we did not measure changes in fitness associated with treatment. This information would have been useful, because fitness change may indicate an overall treatment effect on physical activity.
The present study demonstrates the potential utility of behavioral interventions that modify the relationship between TV viewing and physical activity in obese children. Contingent TV might offer one useful method for modifying this relationship in the home over a short duration of time in obese and sedentary children. Moreover, increased pedaling was highly associated with reduced percent body fat. Studying a larger number of families is necessary to replicate these findings and better determine the extent to which contingent TV can induce body composition changes with and without concomitant dietary modifications. There are currently scarce data on the extent to which specific body composition changes can be induced by particular exercise regimens in pediatric samples.25 The intriguing finding that percent fat was reduced in leg but not other sites awaits replication with larger samples. There is generally limited support for the notion of spot reduction of body fat in the exercise physiology literature, although leg muscle mass apparently can be increased by localized resistance exercises in certain samples.40,41
Contingent TV and comparable interventions might have utility for modifying behavior in school settings given current enthusiasm to increase physical activity in schools.42–44 For certain children (eg, those who avoid group sports), contingent TV might offer a useful physical activity that is consistent with recent recommendations by the American Academy of Pediatrics.44For the practicing pediatrician, the effectiveness of the contingent TV may not be as important as illustrating the power of behavioral principles for modifying sedentary behavior at home. Contingent TVs per se do not necessarily need to be used. Using the same behavioral principles with pedometers, Goldfield et al45 increased obese children's physical activity levels. Pediatricians may be in a unique position to help families reconstruct healthier home environments that reward moderate increases in physical activity that partially substitute for TV viewing. Achieving this may require rigorous assessment of behavioral contingencies within the home environment.
This work was supported in part by Grants RO3DK51931, DK26687 and K08 MH01530 from the National Institutes of Health.
We acknowledge the generous donation of the cycle ergometers used in this study from Lifecycle.
We thank Drs Ross Andersen, Lynn Edlen-Nezin, Jack Wang, and Michelle Porter for their invaluable input on this project.
- Received January 27, 2000.
- Accepted August 3, 2000.
Reprint requests to (D.B.A.) Department of Biostatistics, Ryals Public Health Bldg 327, 1665 University Blvd, University of Alabama at Birmingham, UAB Station, Birmingham, AL 35294-0022. E-mail:
- TV =
- television •
- RPM =
- revolutions per minute •
- V˙o2max =
- maximal oxygen consumption •
- DXA =
- dual-energy x-ray absorptiometry •
- MANOVA =
- multivariate analysis of variance
- Troiano RP,
- Flegal KM
- World Health Organization. Obesity: Preventing and Managing the Global Epidemic. Geneva, Switzerland: World Health Organization; 1998
- Faith MS, Pietrobelli A, Allison DB, Heymsfield SB. Prevention of childhood obesity. In: Bendich A, Deckelbaum RJ, eds. Preventive Nutrition. Totowa, NJ: Humana Press; 1996:471–486
- Faith MS,
- Pietrobelli A,
- Nunez C,
- Heo M,
- Heymsfield SB,
- Allison DB
- Dietz WH,
- Gortmaker SL
- Loncard E,
- Mamelle N,
- Billette A,
- Miginiac M,
- Munoz F,
- Rey S
- Haddock CK,
- Shadish WR,
- Klesges RC,
- Stein RJ
- Jelalian E,
- Saelens BE
- Jason LA, Hanaway LK, Hanaway LK. Remote Control: A Sensible Approach to Kids, TV, and the New Electronic Media. Sarasota, FL: Professional Resources Press; 1997
- Coleman KJ,
- Paluch RA,
- Epstein LH
- Must A,
- Dallal GE,
- Dietz WH
- Derogatis LR. The Brief Symptom Inventory. Minneapolis, MN: National Computer Systems; 1993
- Achenbach TM, Edelbrock C. Manual for the Child Behavior Checklist and Revised Child Behavior Profile. Burlington, VT: University of Vermont, Department of Psychology; 1991
- Rowland TW. Pediatric Laboratory Exercise Testing.Champaigne, IL: Human Kinetics; 1993
- Washington RL,
- Bricker T,
- Alpert BS,
- et al.
- Dunn AL, Blair SN. Exercise prescription. In: Morgan WP, ed.Physical Activity and Mental Health. Washington, DC: Taylor and Francis; 1997:49–62
- Pietrobelli A,
- Formica C,
- Wang ZM,
- Heymsfield SB
- Kushner RF,
- Schoeller DA,
- Fjeld CR,
- Danford L
- Nielson Media Research. 1992–1993 Report on Television. New York, NY: Bun and Bradstreet Corporation; 1993
- Epstein LH. Application of behavioral economic principles to treatment of childhood obesity. In: Allison DB, ed. Obesity Treatment: Establishing Goals, Improving Outcomes, and Reviewing the Research Agenda. New York, NY: Plenum Press; 1995:113–119
- Jeffery DB,
- McLellarn RW,
- Fox DT
- Sipilä S,
- Suominen H
- Gortmaker SL,
- Peterson K,
- Wiecha J,
- et al.
- American Academy of Pediatrics, Committee on Sports Medicine and Fitness and Committee on School Health
- Goldfield GS, Kalakanis LE, Ernst MM, Epstein LH. Open-loop feedback to increase physical activity in obese children. Int J Obes. In press
- Copyright © 2001 American Academy of Pediatrics