Behavioral Mediators of Weight Loss in the SHED-IT Community Randomized Controlled Trial for Overweight and Obese Men
- PDF / 209,537 Bytes
- 7 Pages / 595.276 x 790.866 pts Page_size
- 29 Downloads / 152 Views
BRIEF REPORT
Behavioral Mediators of Weight Loss in the SHED-IT Community Randomized Controlled Trial for Overweight and Obese Men Myles D. Young, BPsyc & David R. Lubans, PhD & Clare E. Collins, PhD & Robin Callister, PhD & Ronald C. Plotnikoff, PhD & Philip J. Morgan, PhD
# The Society of Behavioral Medicine 2014
Abstract Background Little is known about which behavioral strategies are most important to target in weight loss interventions for men. Purpose The aim of the current study was to identify behavioral mediators of weight loss in the male-only Self-Help, Exercise, and Diet using Information Technology (SHEDIT) community weight loss study. Methods A randomized controlled trial with 159 overweight/ obese men [mean (SD) age=47.5 (11.0) years; body mass index=32.7 (3.5) kg/m2] assessed at baseline, 3 months (post-test) and 6 months (follow-up). Results In an intention-to-treat, multiple-mediator model, the significant intervention effect on weight at 6 months (−3.70 kg; p0.05). Therefore, to maximize power, both intervention groups were combined and compared to the control in the current analyses. The mediation analyses were conducted in SPSS Statistics Version 21 (SPSS Inc, Chicago, Illinois, USA) using the INDIRECT Macro [21]. This macro was used to (i) calculate the regression coefficients for the effect of the intervention on the hypothesized mediators (Pathway A), (ii) examine the associations between the mediator variables at 3 months and weight at 6 months, independent of group assignment (Pathway B), and (iii) estimate the total (Pathway C), direct (Pathway C’), and indirect (Pathway AB) intervention effects. All analyses were adjusted for baseline values. This approach is preferred to using change score variables, which are affected by regression to the mean [22]. The macro also generated bias-corrected bootstrapped 95 % asymmetrical confidence intervals around the indirect effect [21]. Significant mediation was established if these confidence intervals did not include zero. Finally, the proportion of the intervention effect attributed to each mediator was calculated by dividing the indirect effect (Pathway AB) by the total effect (Pathway C’+Pathway AB). As recommended in the literature [23], an appropriate temporal sequence was employed to strengthen the evidence for mediation in the current analysis, which investigated whether weight loss at follow-up (6 months) was mediated by post-treatment scores for each hypothesized behavioral mediator at 3 months (for a schematic
representation of the model, see Electronic Supplementary Material (ESM) Figure S1). To adjust for pre-treatment effects, baseline values for weight and all mediator variables were included as covariates in the model. The multiple-mediator model followed an intention-to-treat approach, where missing data were imputed using the expectation maximization procedure in SPSS. This was deemed appropriate as Little’s test did not reject the assumption that the data were missing completely at random (χ2 =161.6, df=144, p=0.15). The amoun
Data Loading...
