Objective: Consideration of adherence is critical in establishing efficacy of a behavioral intervention in the context of a clinical trial. When non-adherence is significant, the "intention-to-treat" approach dilutes the effect of treatments received while the "as-treated" method is subject to bias because of loss of randomized treatment. This report compares three established approaches to evaluating results of a clinical trial testing the efficacy of a nurse-directed walking exercise program on fatigue and other symptoms in women receiving treatment for breast cancer. The approaches were "intention-to-treat", "as-treated", and "instrumental variables with principal stratification". Design: This was a multi-institutional randomized controlled clinical trial stratified by breast cancer treatment modality. Population, Sample, Setting, Years: One hundred nineteen sedentary women with breast cancer receiving outpatient chemotherapy or radiation therapy participated in the study from 1997 through 2001 at four National Cancer Institute designated Cancer Centers and four community cancer centers. Intervention and Outcome Variables: The experimental intervention was an individually prescribed home-based program composed of a brisk, incremental 30-minute walk performed 5 to 6 times a week. The primary outcome was the pretest to posttest change score of fatigue levels. Fatigue was measured by the Piper Fatigue Scale, which has documented reliability and validity. Methods: Following random assignment, subjects in the exercise group maintained an exercise program throughout treatment. The control group received usual care plus attentional control. Data were collected before therapy began and at the end of therapy. Adherence to group assignment was determined from instruments reporting physical activity as well as performance on a 12-Minute Walk Test. Findings: The two groups were similar for baseline demographic and treatment characteristics. Seventy-two percent of subjects randomized to exercise adhered to the exercise prescription. Sixty percent of the control group adhered to the control condition while 40% exercised regularly. Both "intention-to-treat" and "as-treated" data analysis approaches were inadequate to yield an unbiased evaluation of the true effects of the intervention. In consideration of the adherence issue, the results were analyzed using the method of "instrumental variables with principal stratification" (IV/PS). The IV/PS approach is an accepted reputable method in statistics but less known in other sciences. This method demonstrated that regular exercise has clinically and statistically significant effects (p=0.03) on fatigue levels. Conclusions: A moderate home-based walking exercise program can effectively mitigate the high levels of fatigue common during breast cancer treatment. Nursing research includes the testing of behavioral interventions (such as exercise, diet, and medication administration) in which non-adherence may include cross-over from one randomized group to the other. Approaches to data analysis must include consideration of adherence in order to determine the true effects of receiving the treatment. Implications for Nursing Practice: Nurse-prescribed and monitored walking exercise is an effective, practical self-care activity to reduce symptoms during breast cancer treatment.
Back to Studies in Research and Data Management
Back to The Advancing Nursing Practice Excellence: State of the Science