Meta-Analysis of Anthropometric Outcomes of Motivational Physical Activity

Interventions

Jo-Ana D. Chase MN APN-BCTodd Ruppar PhD RN GCNS-BC

Lorraine J. Phillips PhD RNVicki S. Conn PhD RN FAAN

University of Missouri Sinclair School of Nursing

Acknowledgements

• Parent study and funding: Meta-analysis of interventions to increase exercise/physical activity (R01NR009656, Dr. Vicki Conn – Principal Investigator)

Background

• Prevalence of obesity increasing• Obesity linked with multiple chronic illnesses (Guh,

et al, 2009; Suzuki, Orsini, Safi, Key, & Wolk, 2009)• Inadequate physical activity (PA) contributes to

epidemic of obesity (Rossner, et al., 2008; Wareham, van Slujs, & Ekelund, 2005)

• Many interventions designed to motivate subjects to increase PA have also examined anthropometric outcomes

Background

• Prior meta-analyses of PA anthropometric outcomes• Limited with no moderator analysis• No separation of outcomes for diet or PA

interventions• Most comprehensive meta-analysis of PA on

anthropometric outcomes – Ballor and Keesey, 1991

Research Questions

• What is the overall effect of interventions designed to motivate subjects to increase PA on anthropometric outcomes?

• Do anthropometric effects of these interventions vary depending on intervention, sample, or design attributes?

Methods

• Comprehensive search motivational PA intervention studies reporting anthropometric outcomes (e.g., BMI, percent body fat, central obesity) in healthy adult subjects

• “Motivational PA intervention” = any deliberate intervention intended to motivate subjects to increase PA behavior, excluding supervised exercise sessions

Methods – Data Extraction

• Coding frame developed from prior, related research and pilot tested• Study characteristics (e.g.,

published/unpublished, funding)• Sample characteristics (e.g., age, gender)• Intervention characteristics (e.g., dose,

delivery)• Effect size information

Methods – Data Extraction

• Studies coded by two independent and extensively trained coders

• Coders compared all variables to reach consensus

• All effect size data independently verified by doctorally prepared researcher

Methods - Analysis

• Standardized mean difference effect sizes (ES, d) under random-effects model

• Studies weighted by inverse of the variance• Heterogeneity examined: Q statistic and I2

• Exploratory moderator analyses – meta-analytic analogues of ANOVA (dichotomous moderators) and regression (continuous moderators)

Pooled standard deviation

d = meantreatment - meancontrol

Study Characteristics

• Reviewed 54,642 potentially eligible reports• ES calculated from data for 94,711 subjects• Two group comparisons:

• 76 comparisons• 94,711 subjects

• Single group, treatment pre-post comparisons• 436 comparisons• 56,258 subjects

Sample Characteristics

• Sample sizes• Two group: median 56 subjects• Single group: median 52 subjects

• Percent female: median 81%• Percent minority: median 15%• Baseline BMI: median 27-30 kg/m2

Intervention Characteristics

• Number of intervention strategies: median 3• Total number of intervention minutes: median

720 minutes• Recommended minutes of PA per session:

median 30 minutes• Recommended minutes of PA per week:

median 140 minutes

Results – Anthropometric Outcomes

Comparison Type k ES p I2

Treatment vs. control at outcome 76 0.09 <.001 .31

Treatment pre-post vs. control pre-post

67 0.10 <.001 .84

Treatment pre-post 436 0.20 <.001 .97

Control pre-post 53 -0.03 <.05 .85

k=number of comparisonsES=effect sizep=test of statistical significance for effect sizeI2=index of heterogeneity beyond within-study sampling error

Effect Size Meanings

• Converted ES to original metrics – weight (kg) and BMI• Mean difference of 1.3 kg between treatment and

control subjects at outcome• BMI difference of 0.4 between treatment and

control subjects at outcome (ex. 27 vs. 27.4)

Results – Moderator Analyses

• Larger ESs for interventions using transtheoretical model (TTM) (d=0.21) vs. social cognitive theory (SCT) (d=0.07)

• Studies with mean ages in 30s and early 60s demonstrated higher ESs than studies with other age groups

• Studies recommending walking were not more effective in improving anthropometric outcomes than studies recommending other forms of exercise

Limitations• Study specific

• Anthropometric measurement error• Low reporting of ethnic diversity among samples• Long-term results not often examined• Study quality and treatment fidelity

• Meta-analysis specific• Exclusion of studies based on insufficient

information• Findings observational• Moderator analyses – hypothesis generating

Discussion

• Findings support significant improvements in anthropometric outcomes associated with motivational PA interventions

• Future research to directly compare TTM-based to SCT-based interventions

• Future research to test interventions recommending varied forms of PA (not just walking)

References• Ballor, D. L., & Keesey, R. E. (1991). A meta-analysis of the factors affecting

exercise-induced changes in body mass, fat mass and fat-free mass in males and females. International Journal of Obesity, 15(11), 717-726.

• Guh, D. P., Zhang, W., Bansback, N., Amarsi, Z., Birmingham, C. L., & Anis, A. H. (2009). The incidence of co-morbidities related to obesity and overweight: a systematic review and meta-analysis. BMC Public Health, 9, 88.

• Rossner, S., Hammarstrand, M., Hemmingsson, E., Neovius, M., & Johansson, K. (2008). Long-term weight loss and weight-loss maintenance strategies. Obesity Reviews, 9(6), 624-630.

• Suzuki, R., Orsini, N., Saji, S., Key, T. J., & Wolk, A. (2009). Body weight and incidence of breast cancer defined by estrogen and progesterone receptor status--a meta-analysis. International Journal of Cancer, 124(3), 698-712.

• Wareham, N. J., van Sluijs, E. M. F., & Ekelund, U. (2005). Physical activity and obesity prevention: a review of the current evidence.[erratum appears in Proc Nutr Soc. 2005 Nov;64(4):581-4]. Proceedings of the Nutrition Society, 64(2), 229-247.