Research Article: Contamination by an Active Control Condition in a Randomized Exercise Trial

Date Published: October 10, 2016

Publisher: Public Library of Science

Author(s): Diane K. Ehlers, Jason Fanning, Elizabeth A. Awick, Arthur F. Kramer, Edward McAuley, Maciej Buchowski.


Contamination is commonly overlooked in randomized trials. The present study examined contamination (minutes of aerobic activity outside of exercise sessions) within an active control condition in a 6-month randomized exercise trial for older adults. We hypothesized that outside aerobic activity would be greater in the control condition compared to the intervention conditions. Participants (mean age = 65.06 years, 66.2% female) were randomly assigned to: Dance (n = 50), Walking, (n = 108), or Strength/Stretching/Stability (SSS; n = 48). Dance and Walking represented the experimental conditions and SSS the control condition. Participants attended exercise sessions three times weekly for 24 weeks. Participants recorded their physical activity outside of class on a weekly home log. Group assignment and covariates (age, gender, body mass index, exercise session intensity and enjoyment, and program adherence) were examined as predictors of weekly aerobic activity outside of exercise sessions. Participants who returned zero home logs were removed from the dataset (final N = 195). Out-of-class aerobic activity was lowest in the Walking group. Significant effects of gender, group, enjoyment, and intensity on out-of-class weekly aerobic activity were observed, all p<0.003. Higher perceived enjoyment of exercise sessions was associated with more out-of-class aerobic activity, while higher perceived intensity was associated with less out-of-class aerobic activity. A group x intensity interaction, p = 0.002, indicated that group differences in out-of-class aerobic activity were evident only among those with lower intensity perceptions. Walkers may have perceived exercise sessions as sufficient weekly exercise, while the Dance and SSS groups may have perceived the sessions as necessary, but insufficient. The lower aerobic intensity Dancers attributed to exercise sessions and non-aerobic nature of SSS may partially explain contamination observed in this study. Further examination of contamination in randomized controlled exercise trials is critically needed.

Partial Text

Although most researchers aim to maximize adherence among participants enrolled in intervention trials, minimizing contamination between conditions is an often overlooked problem in health behavior research [1,2]. Adherence refers to the degree to which the participant fulfills or completes intervention activities and is typically reported in published randomized controlled trials (RCTs). Contamination, on the other hand, reflects the extent to which participants in a control condition adopt the experimental treatment [1]. For example, in a study testing the effects of aerobic exercise on a particular health outcome, adoption of aerobic activity by the control condition may serve as a point of contamination. Designing orthogonal intervention and control groups in health behavior and psychological research is often challenging due to potential placebo effects, Hawthorne effects, attentional differences, treatment preferences, and differential outcome expectations between groups [3–7]. These biases may cause researchers to commit Type I or II errors by way of incorrectly attributing effects resulting from these biases to the active intervention or failing to observe significant effects due to control participants’ adoption of intervention-targeted behaviors [3]. Health behavior RCTs, such as exercise trials, may be particularly susceptible to contamination because participants are often unblinded to group assignment; characteristics of the groups, such as the group facilitator or social environment, may differ; and prevention of non-study activity behaviors are ethically beyond the control of the researcher [4,7–9].

The purpose of this study was to examine the contamination by an active, non-aerobic comparison group in a randomized exercise trial. Specifically, we examined group differences in self-reported aerobic activity outside of scheduled exercise sessions. Out-of-class aerobic activity decreased in the Walking group across the intervention, but was maintained in Dance and SSS (Fig 2). Major findings suggest that exercise mode and perceived intensity may have contributed to participants’ weekly activity outside of exercise sessions. The Walking group may have perceived the three-times weekly walking sessions as sufficient, while the Dance and SSS groups may have perceived the three-times weekly exercise sessions as a necessary, but insufficient amount of weekly exercise, particularly if perceptions of intensity were low. While enjoyment of exercise sessions emerged as a significant predictor of out-of-class aerobic activity across groups, intensity was the stronger predictor and may have been responsible for group differences observed. Traditional pre-post measures of physical activity may not be adequate for examining contamination effects. The weekly home logs, on the other hand, may provide more comprehensive insight into activity behaviors across conditions.

In summary, the findings contribute to the scientific discussion regarding the balance between rigor and ethics in designing appropriate comparison groups. Research on the contamination effects of control groups is noticeably absent from the literature, yet represents an important area of study for advancing our knowledge about physical activity’s contribution to a variety of health outcomes. Findings suggest that moderate-to-vigorous intensity aerobic interventions may result in perceptions of sufficiency, leading participants to engage in less aerobic activity outside of scheduled sessions. Participants assigned to non-aerobic or lower intensity aerobic interventions, on the other hand, may view exercise sessions as necessary, but not sufficient. Further examination of contamination effects in randomized controlled exercise trials is critically needed.

This study was approved by the University of Illinois at Urbana-Champaign Institutional Review Board. All procedures performed were in accordance with the ethical standards of the institutional review board and with the 1964 Helsinki declaration and its later amendments.