Table 4.
Effects and moderators of the effects of exercise on self-reported cognitive functioning
| Self-reported cognitive functioning | z-score | |
|---|---|---|
| β (95% confidence interval) | p-value LRT | |
| Overall exercise effect (n=1987) | −0.09 (−0.16; −0.02) | |
| Exercise effect after treatment (n=745) | −0.13 (−0.24; −0.02) | |
| Exercise effect during treatment (n=1162) | −0.08 (−0.18; 0.02) | |
| Sociodemographic moderators | ||
| Age | No significant interaction | 0.17 |
| Education level | No significant interaction | 0.73 |
| Gender | No significant interaction | 0.32 |
| Clinical moderators | ||
| Chemotherapy | No significant interaction | 0.43 |
| Hormone therapy (breast cancer) | No significant interaction | 0.12 |
| Cancer type | No significant interaction | 0.32 |
| Baseline levels of patient-reported outcomes | ||
| Baseline fatigue | No significant interaction | 0.38 |
| Baseline cognitive functioning | No significant interaction | 0.60 |
| Depression* | No significant interaction | 1.00 |
| Anxietya,* | 0.08 | |
| Yes (n=325) | 0.07 (−0.12; 0.26) | |
| No (n=1314) | −0.10 (−0.19; −0.02) | |
| Intervention moderators | ||
| Delivery modeb | 0.97 | |
| Unsupervised | Reference | |
| Supervised | −0.004 (−0.15; 0.14) | |
| Intervention length | 0.07 | |
| =<12 weeks | −0.14 (−0.25; −0.04) | |
| 12–24 weeks | 0.01 (−0.13; 0.15) | |
| >24 weeks | −0.18 (−0.32; −0.02) | |
aOriginally, anxiety was measured on a continuous scale. The score was dichotomized using cutoff scores from literature: HADS-A >9 [52], or using the mean: BSI and SCL-90
bInteraction testing is not applicable; therefore, differences between subgroups are reported. LRT of the model including the intervention characteristic vs the main model is presented
LRT likelihood ratio test
The respective z-scores are calculated using the MFI, CIS, FAQ and EORTC QLQ C30 questionnaire
A negative z-score indicates an improvement in cognitive functioning
*Excluding two studies [31, 35] which did not measure depression or anxiety