We have carefully read the comments by Goedendorp et al. and noted
Goedendorp's argument that the effect on fatigue of our exercise
intervention was smaller than evidenced. We estimated a 6.6-point mean
difference in fatigue between the control and the intervention group,
using a general linear regression model adjusted for a number of possible
confounders, including baseline values. This statistical method
strengthens the value of the mean difference estimate by considering
possible confounders. In our opinion, the suggestion by Goedendorp of a
proposed observed effect in mean difference of 3.1 point therefore is too
simplistic a calculation. In order to interpret the significance of
changes in patient related outcome and to compare our findings with
existing meta analyses in cancer and exercise, clinically important
changes were estimated as effect sizes using Cohen’s guidelines [1,2]. The
estimated 6.6-point mean difference is equivalent to an effect size (ES)
.33 that suggests a small to medium clinically important change in
fatigue.
Goedendorp et al. refers to a series of studies that reject the
effect of physical activity on reducing fatigue in patients with Chronic
Fatigue Syndrome (CFS). We are forced to reject the relevance of drawing
in these studies to compare against the findings of our study, since
Goedendorp et al research CFS in a heterogeneous sample of patients and
not fatigue in cancer patients while undergoing chemotherapy, as in our
study. Furthermore, the CFS studies referred to by Goedendorp do not use
exercise as an intervention component.
Goedendorp argues that other interventions than exercise may
influence the patient's feelings of fatigue. We agree, and our multimodal
intervention design, that includes relaxation training, body awareness
training and massage, confirms our presumption that components other than
high intensity exercise may contribute to the improvement in cancer
patients' physical and emotional well-being. Our intervention was
therefore offered as a package and must be viewed as an entity within
which each component plays a role in the outcome.
1. Cohen J. Statistical power analysis for the behavioral sciences.
2nd rev ed. Hillsdale: Lawrence Erlbaum Associates, 1988.
2. Nakagawa S, Cuthill IC. Effect size, confidence interval and
statistical significance: a practical guide for biologists. Biol Rev Camb
Philos Soc 2007;82:591-605.
Competing interests:
None declared
Competing interests:
No competing interests
04 December 2009
Lis Adamsen
Professor Clinical Nursing
Tom Møller, Maria Stage and Mikael Rørth
University Hospitals Centre for Nursing and Care Research, Cph. Univ. Hospital, DK-2100 Copenhagen
Rapid Response:
Authors' reply to comments by Goedendorp et al.
We have carefully read the comments by Goedendorp et al. and noted
Goedendorp's argument that the effect on fatigue of our exercise
intervention was smaller than evidenced. We estimated a 6.6-point mean
difference in fatigue between the control and the intervention group,
using a general linear regression model adjusted for a number of possible
confounders, including baseline values. This statistical method
strengthens the value of the mean difference estimate by considering
possible confounders. In our opinion, the suggestion by Goedendorp of a
proposed observed effect in mean difference of 3.1 point therefore is too
simplistic a calculation. In order to interpret the significance of
changes in patient related outcome and to compare our findings with
existing meta analyses in cancer and exercise, clinically important
changes were estimated as effect sizes using Cohen’s guidelines [1,2]. The
estimated 6.6-point mean difference is equivalent to an effect size (ES)
.33 that suggests a small to medium clinically important change in
fatigue.
Goedendorp et al. refers to a series of studies that reject the
effect of physical activity on reducing fatigue in patients with Chronic
Fatigue Syndrome (CFS). We are forced to reject the relevance of drawing
in these studies to compare against the findings of our study, since
Goedendorp et al research CFS in a heterogeneous sample of patients and
not fatigue in cancer patients while undergoing chemotherapy, as in our
study. Furthermore, the CFS studies referred to by Goedendorp do not use
exercise as an intervention component.
Goedendorp argues that other interventions than exercise may
influence the patient's feelings of fatigue. We agree, and our multimodal
intervention design, that includes relaxation training, body awareness
training and massage, confirms our presumption that components other than
high intensity exercise may contribute to the improvement in cancer
patients' physical and emotional well-being. Our intervention was
therefore offered as a package and must be viewed as an entity within
which each component plays a role in the outcome.
1. Cohen J. Statistical power analysis for the behavioral sciences.
2nd rev ed. Hillsdale: Lawrence Erlbaum Associates, 1988.
2. Nakagawa S, Cuthill IC. Effect size, confidence interval and
statistical significance: a practical guide for biologists. Biol Rev Camb
Philos Soc 2007;82:591-605.
Competing interests:
None declared
Competing interests: No competing interests