Conn, V. S., Hafdahl, A. R., Porock, D. C., McDaniel, R., & Nielsen, P. J. (2006). A meta-analysis of exercise interventions among people treated for cancer. Supportive Care in Cancer, 14, 699–712.

Databases searched were MEDLINE, CANCERLIT, Cochrane Central Register of Controlled Trials (CENTRAL), Dissertation Abstracts, PsycINFO, SPORTDiscus, HealthSTAR, Clinical Evidence, and CINAHL through 2002. The authors also searched the National Institutes of Health (NIH) database of funded studies from 1986 through 2002 and conducted hand searches of selected journals.

Thirty primary study reports (24 in the published literature and the remainder in unpublished dissertations and presentation papers) contained sufficient information to be included in the quantitative analysis. Thirteen studies were designed as single-group pre/post research; the remainder compared at least two groups of patients. Only one two-group study did not randomly assign patients to study arms. Comparisons groups were most often described as having received usual care. Outcomes were quality of life, physical functional ability, fatigue, symptoms other than fatigue, mood, body composition, and exercise behavior. 

Twenty-one of 30 studies tested supervised exercise interventions rather than home-based exercise. Supervised exercise was most often scheduled three times per week, and in 16 of the 21 studies of supervised exercise, the exercise program lasted longer than 10 weeks. Supervised exercise generally included aerobic activity (e.g., walking and cycling) and less often included resistance or flexibility exercise. Mixed types of aerobic exercise were used in several studies. The exercise intervention in most studies was of moderate intensity to achieve approximately 30% to 70% of maximum oxygen consumption.

• In 13 studies, the sample was comprised exclusively of women with breast cancer, and, in five other studies, women with breast cancer comprised more than half of the patients.
• Patients with other cancers were included less commonly, and few studies included patients with more advanced cancer, older patients, or minority patients.
• Sample sizes ranged from five to 155 patients.
• Age ranged from 31 to 71 years.

The effect size estimate for the outcome of fatigue in the two-group comparisons was small and not statistically significant (standardized mean difference [SMD] = 0.11). However, the effect size estimate for the outcome of fatigue in the single group comparisons of pre- and posttest was larger (SMD = 0.27) and remained statistically significant under both the assumption that there was no correlation between participants’ pre- and posttest scores and that pre- and posttest scores were strongly correlated (r = 0.80).

Effect sizes among only control group participants were very small but negative; this observation may lend some support to the validity of meta-analytic findings from single-group designs.

The small, nonsignificant effect size for exercise on the outcome of fatigue may have occurred due to heterogeneity in the exercise characteristics and intervention dose, varied samples, diverse measures of outcomes, and variable outcome assessment timing.