Table 3 Summary of findings for the main comparisons

Bantum 2014 [14]

Brief Fatigue Inventory (BFI)

303

156

147

6 months

p = 0.56 Effect size = 0.17 (Calculated by taking the differences of the means at 6 months predicted from the model, including adjustment factors, divided by the standard deviation for the difference computed from the within and between subject variance components.)

Control group,

• Baseline (n = 176); mean (95% CI) = 40.8 (38.9–42.8)

• Month 6 (n = 156); mean (95% CI) = 40.7 (38.7–42.8)

Intervention group

• Baseline (n = 176); mean (95% CI) = 39.0 (37.0–40.9)

• Month 6 (n = 147); mean (95% CI) = 36.4 (34.2–38.5)

Bennett 2007 [15]

Schwartz Cancer Fatigue Scale

56

28

28

6 months

On average, the level of fatigue status for all participants was 15.20 at baseline and declined 4.22 points (27%) across the study.

Group × Time interaction for fatigue was significant [Λ =0.78, F(2,37) = 5.24, p = 0.010]. However, inspection of the graph showed this was an artifact of 3-month measures, whereas values at baseline and at 6 months showed no significant differences between groups, leading to the conclusion that the significant effect of the interaction was the result of measurement error.

Blaes 2016 [16]

Functional Assessment in cancer Therapy-Fatigue ( FACT-F)

42

28

14

4 months

There was an improvement in fatigue in both groups with time. Mean improvement from baseline to 4 months was 6.8 for the MBCR group and 1.3 for controls (p = 0.19).

There was no statistically significant difference in improvement in fatigue for two groups.

Bower 2015 [17]

Fatigue Symptom Inventory

71

39

32

3 months

Mindfulness led to significant improvements in fatigue (p = 0.007), from pre- to post-intervention.

No group differences in change from baseline to 3-month follow-up p = 0.57

Bruggeman-Everts 2017 [18]

Checklist Individual Strength - Fatigue Severity [CIS-FS] subscale

167

55

112

9 weeks

AAF = eMBCT = psycho-educationχ2(4)=27.63, p < .001

AAF = psycho-educationχ2(2)=28.28, p < .001

eMBCT = psycho-educationχ2(2)=10.89, p = .004

AAF = eMBCTχ2(2)=2.19, p = .34

Multiple group latent growth curve analysis, corrected for individual time between assessments, showed that fatigue severity decreased significantly more in the AAF and eMBCT groups compared to the psychoeducational group.

Carlson 2013 (2016) [19, 45]

POMS

271

113

158

6 and 12 months later.

Group-by-time effect at intervention (6months): p = 0.001

95% CI − 0.45 [− 0.70; − 0.20]

Group-by-time effect at follow-up (12 months) p = 0.76

Dirksen 2008 [20]

Profile of Mood States Fatigue/Inertia Subscale (POMSF/I)

72

34

38

2 weeks

Statistically significant pre- to post-treatment change (p < 0.05).

From pre- to post-treatment, the CBT-I group improved on fatigue. Statistically significant interaction effects were found for fatigue At post-treatment, a trend was noted towards lower fatigue [t(70) = 1·87, p = 0·07].

Dodds 2015 [21]

Medical Outcomes Study Short Form 12-Item HealthSurvey (SF-12)

28

16

12

4 weeks

Improvement in fatigue/vitality From baseline to study week 8 = 5.5,

95% CI [1.5; 9.6];

1-month FU 0.3

95% CI [−4.2; 4.9] no significant differences at the 4- week follow-up.

Dolbeault 2009 [22]

POMSF/I and EORTC Fatigue

167

81

86

6 months

Comparison of change scores between randomisation arms (Group: n=81; Control: n=87)

POMS fatigue

• Group: E1 Mean (SD) 10.01 (7.38) ; E3 Mean (SD) 6.86 (5.58) ; Intra-subject p = -0.069 Eta²= 0.02

• Control: E1 Mean (SD) 8.78 (6.85); E3 Mean (SD) 8.87 (6.84) Inter-subject p = 0.370 Eta²= 0.01

• Time X group p = 0.000 Eta²= 0.07

EORTC Fatigue

• Group: E1 Mean (SD) 2.24 (0.81) ; E3 Mean (SD) 2.08 (0.73) Intra-subject p = 0.834 Eta² = 0.00

• Control E1 Mean (SD) 2.09 (0.68) ; E3 Mean (SD) 2.14 (0.77)

• Inter-subject p = 0.408 Eta² = 0.00

• Time X group p = 0.036 Eta² = 0.03

A greater reduction of negative affects and improvement in positive affects and in quality of life functional or symptom scales were observed in the TG compared with the CG. This concerned the POMS fatigue (7% of the variance explained by the model including the time/group interaction term) and the EORTC QLQ-C30 fatigue (3%).

Espie 2008 [23]

FSI

150

100

50

6 months

p < 0.001 (Standardized Effect =− 0.82)

CBT participants had reduced symptoms of fatigue relative to TAU.

FSI Interference

Post-Treatment

• Standardized Effect - 0.81

• 95% CI − 1.20 to-0.42

• p < 0.001

6-Month follow-up

• Standardized Effect − 0. 82

• 95% CI − 1.22 to − 0.42

• p < 0.001

Ferguson 2016 [24]

Functional Assessment of Chronic Illness Therapy-Fatigue [FACIT-F]

47

27

20

2 months

Memory and Attention Adaptation Training (MAAT) and Supportive Therapy (ST) participants did not differ with regard to fatigue (FACIT-F) at the post-treatment (F (1,28), 0.072; p = 0.79) or 2-month ((F (1,28), 2.35; p = 0 .14). The Cohen’s d effect sizes for, fatigue at the 2-month follow-up time point suggested that MAAT participants demonstrated sustained clinical gains compared with ST participants (0.46)

Fillion 2008 [25]

Multidimensional Fatigue Inventory

87

44

43

3 months

Marginal Group x Time interaction effects: p = 0 .07; Cohen d = 0.36

Significant time main effects: p = 0 .0001; Cohen d = 0.69

Significant group main effects: p = 0 .03; Cohen d = 0.49

Results showed that participants in the intervention group

showed greater improvement in fatigue.

Foster 2016 [26]

Brief Fatigue Inventory (BFI)

159

83

76

12 weeks

T1 Group effect (95 % CI) 0.514 (− 0.084, 1.112) p = 0.09

T2 Group effect (95 % CI) 0.106 (− 0.427, 0.638) p = 0.70

Freeman 2015 [27]

FACIT-Fatigue and Scale (FACIT-F, version 4)

118

71

47

3 months

Group effect p value = 0.002

Time effect p value= 0.084

Group × time effect p value = 0.321

The Bonferroni method was used to correct for multiple comparisons, and alpha was adjusted to 0.01. Linear multilevel modelling analyses revealed less fatigue, cognitive dysfunction, and sleep disturbance for Live Delivery and Telephone Delivery compared with WL across the follow-up (p’s < 0.01). Changes in fatigue, cognitive dysfunction, sleep disturbance, and health-related and breast cancer-related QOL were clinically significant. There were no differences between LD and TD.

Gielissen 2006 [28]

Fatigue severity subscale of the CIS

98

50

48

6 months

Patients in the intervention condition reported a significantly greater decrease than patients in the waiting list condition in fatigue severity (difference, 13.3; 95% CI, 8.6 to 18.1)

Heckler 2016 [29]

Brief Fatigue Inventory (BFI)/ FACIT-F

96

47

49

7 weeks (post intervention)

CBT and placebo p = 0.0005 (95% CI) [− 2.22, − 0.74]

CBT and placebo p ≤ 0.0001 (95% CI) [5.57, 12.90]

CBT-I effect (95% CI) for BFI was − 1.00 (− 1.64, − 0.37), p = 0.0024, meaning that CBT-I led to a mean change one unit less than no CBT-I.

The CBT-I effect (95 % CI) for FACIT-Fatigue was 7.16 (3.68, 10.64), p < 0.0001, meaning that CBT-I led to a mean change seven units higher than no CBT-I.

No statistically significant change between post-intervention and follow-up; p = 0.294 (BFI), p = 0.145 (FACIT-Fatigue).

Hoffman 2012 [30]

pOMSF/I

214

103

111

12–14 weeks

There were statistically significant differences between treatment groups for POMS fatigue p = 0.002 [8 weeks only]

Difference Between Groups at T2 adjusted for baseline mean = − 2.68; 95% CI = [− 4.31 to − 1.04]

Difference between groups at T3 adjusted for baseline mean = − 1.84 95% CI = [− 3.45 to − 0.22]

Interaction time X treatment group, P .324

Johns 2015 [49]

Fatigue Symptom Inventory

35

18

17

1 month

Significantly greater improvements in fatigue interference than wait-list controls. The magnitude of the effect of MBSR on this and other fatigue outcomes including fatigue severity and vitality was large at the end of the intervention and 1 month later. improvements in all symptoms were maintained for at least 6 months beyond the completion of the MBSR course for both groups after their respective courses.

T2

FSI interference

p* ≤ 0.001 Pooled SD = 1.73 Effect size =− 1.43 95% CI effect size = [1.96, − 0.90]

FSI severity

p* ≤ 0.001 Pooled SD = 1.64 Effect size =− 1.55 95% CI effect size = [− 2.09, − 1.01]

T3

FSI interference

p* ≤ 0.001 Pooled SD = 2.01 Effect size=− 1.34 95% CI effect size = [1.88, − 0.81]

FSI severity

p* ≤ 0.001 Pooled SD = 1.51 Effect size =− 1.54 95% CI effect size = [− 2.10, − 0.97]

Lengacher 2012 [31]

Symptom Inventory (MDASI)

84

41

43

6 week

p < 0.5

P (between-group post-assessment) p = 0.05

At post-intervention, the MBSR(BC) group showed greater improvement across symptoms, and especially symptom interference items, compared to the control group. For the MBSR(BC) group, statistically-significant reductions (p < 0.01) were observed for fatigue.

Matthews 2014 [32]

Piper Fatigue Scale

56

30

26

6 week

p = 0.76 d = 0.2

No group differences in improvement were noted relative to fatigue.

Prinsen 2013 [33]

Checklist Individual Strength (CIS-fatigue)

37

23

14

6 months

CBT resulted in a significantly larger decrease in fatigue severity compared to a period of waiting for therapy.

After 6 months of follow-up, patients who underwent CBT, with a mean of 12.0±5.0 individual sessions, showed a significantly larger change in fatigue scores than patients in the waiting list group (p < 0.001, respectively − 49.0 ± 23.0% and − 16.4 ± 25.0%).

Baseline to follow-up (within group) p < 0.001 p = 0.022

Reeves 2017 [34]

FACIT

90

45

45

6 months

Only the intervention arm showed significantly improved

Fatigue- Mean change (95% CI)= 3.0 (0.7, 5.3) p < 0.01

Intervention – usual care- No statistically significant intervention effects were observed

Mean difference (95% CI) = 1.1 (− 2.4, 4.5)

p = 0.527

Reich 2017/ Lengacher 2016 [35, 46]

Fatigue Symptom Inventory

303

155

148

12 weeks

MBSR(BC) demonstrated greater symptom improvement in fatigue (severity and interference; p < 0.01).

Effect sizes (Cohen’s d) were between 0.27 and 0.23. A majority of improvements in fatigue occurred during the MBSR(BC) training, with little change occurring during the follow-up period (6 to 12 weeks). Fatigue—severity (FSI) p = 0.002T2 week d = 0.33 95% CI [0.13 to 0.54] T3 week d = 0.27 95% CI 12 0.07 to 0.47 Fatigue—interference (FSI) p = 0 .006T2 week d = 0.3 95% CI [0.10 to 0.51 ]T3 week d = 0.23 95% CI [0.02 to 0.43]

Reif 2013 [47]

Fatigue Assessment Questionnaire (FAQ) and Fatigue subscale of the EORTC-QLQ-C30

234

120

114

6 months

FAQ : Significant reduction in intervention group: (F = 76.510, p < 0.001, η2 = 0.248). The control group showed almost no change in CRF levels over time. In the repeated measures ANOVA, this difference was statistically significant for the group by time interaction (F = 76.51, p < 0.001). The partial η2of 0.248 indicates a large effect.

QLQ-C30 fatigue subscale: the IG showed a reduction from 75.37 (19.39) to 40.74 (30.60) while the values in the CG remained about the same (F = 57.837, partial η2 = 0.2, p < 0.001). This finding confirms the results of the FAQ.

Ritterband 2012 [36]

Multidimensional Fatigue Symptom Inventory- Short Form (MFSI-SF)

28

14

14

9 weeks

p < 0.01

Overall adjusted ES (d) = 1.16

A significant group × time interaction was found for the overall measure of fatigue, MFSI-SF (F_1,26 = 13.88, p < 0.01). Participants in the Internet group had significantly improved fatigue scores from 22.86 to 9.50 (t(13) = 3.63, p < 0.01); control participants’ scores did not improve over time, changing from 13.71 to 19.79 (t(13) = − 1.64, p = 0.12). Several MFSI-SF subscales also had significant group × time interactions, including general fatigue (F_1,26 = 9.46, p < 0.01), mental fatigue (F_1,26 = .65, p < 0.01), and vigor (F_1,26 = 14.79, p < 0.01), with Internet participants showing improvements compared with control participants in all cases. Although some subscales lacked significant group × time interactions (physical fatigue, p = 0.11; emotional fatigue, p = 0.08), adjusted ES for the fatigue variables ranged from a low of 0.47 to a high of 1.63, indicating a SHUTi treatment effect for fatigue.

Rogers 2017 [37]

Fatigue Symptom Inventory

222

110

112

3 months

BEAT Cancer significantly reduced fatigue intensity at both time points (mean between group difference [M] = − 0.61; 95% CI = − 1.04 to − 0.19; effect size [d] = − 0.32; p = .004 at M3 and M = − 0.46; 95% CI − 0.89 to − 0.03; d = − 0.26; p = .038 at M6).

Significant and greater reductions in fatigue interference

occurred (M = − 0.84; 95% CI = − 1.26 to − 0.43; d = − 0.40;

p < .001 at M3 and − 0.66; CI − 1.08 to − 0.24; d = − 0.35; p = .002 at M6).

Sandler 2017 [38]

46

22

24

24 weeks

Fatigue severity improved in all subjects from a mean of 5.2 (− 3.1) at baseline to 3.9 (− 2.8) at 12 weeks, suggesting a natural history of improvement. Clinically significant improvement was observed in 7 of 22 subjects in the intervention group compared with 2 of 24 in the education group (p < 0.05)

The whole cohort reported improvements in fatigue scores between baseline and 12 weeks (Mdiff = − 1.27; 95% CI − 2.52 to − 0.03; p < 0.05) and 24 weeks (Mdiff = − 1.51; 95% CI − 2.84 to − 0.18; p < 0.05).

Change scores differed significantly in favour of the intervention (M = 2.55, SD = 3.77; t(36) = − 2.56; p < 0.05) at 12 weeks in comparison to the education arm (M = 0.10; SD = 2.55) but not at follow up (Mdiff = 1.56; 95% CI − 3.77 to 0.48; p = 0.13).

These groupwise changes indicate an effect size in the CBT/GET group of d = 0.79, compared with d = 0.04 in the education arm.

Savard 2005 [39]

Multidimensional Fatigue Inventory (MFI)

57

27

30

12 months

Pooled data revealed significant differences between pre- and post-treatment on fatigue (F1,158 = 11.70; p < .001), No significant difference was detected between post-treatment and the follow-up evaluations.

Therapeutic effects were well maintained up to 12 months after the intervention and generally were clinically significant.

Pooled data

(n = 57)

3-month follow-up : adjusted mean= 2.33; 95% CI = 2.15 to 2.51

6-month follow-up: adjusted mean = 2.25; 95% CI = 2.07 to 2.43

12-month follow-up: adjusted mean = 2.18; 95% CI = 1.98 to 2.38

Van Der Lee 2012 [40]

Multidimensional Fatigue Inventory (MFI)- General fatigue

83

59

24

6 months

p < 0.001

At post-treatment measurement the proportion of clinically improved participants was 30%, versus 4% in the waiting list condition (Χ² (1) = 56.71; p = 0.007).

The mean fatigue severity score at post-measurement was significantly lower in the intervention group (95%CI = 33.2–37.9) than in the waiting list group (95% CI = 40.0–47.4) controlled for pre-treatment level of fatigue. The effect size for fatigue is 0.74 (d = (mean post intervention–mean post control)/pooled SD).

The treatment effect was maintained at 6-month follow-up. At follow up 39% of the participants in the intervention group

showed clinically relevant improvement in fatigue severity.

Van Weert 2010 [41]

Multidimensional Fatigue Inventory (MFI)- General fatigue

209

76

133

12 weeks

In comparison with the WLC group, the PT group showed more reduction in 4 domains of fatigue, whereas the PT+CBT group showed more reduction in one domain only. Finally, the results showed that physical training combined with CBT and physical training alone were equally effective in reducing fatigue. Thus, CBT did not seem to contribute additional positive effects on fatigue to the benefits of physical training.

PT + CBT (WLC = Reference) between-group change

General fatigue (95% CI) = − 1.3 (− 3.1 to 0.4)

Physical fatigue (95% CI) = − 2.7 (− 4.5 to − 1.0) p < 0.01.

Mental fatigue (95% CI) = − 0.5 (− 2.3 to 1.2)

Reduced motivation (95% CI) = − 0.6 (− 2.1 to 1.0)

Reduced activation (95% CI) = − 0.9 (− 2.6 to 0.8)

Willems 2017 [48]

Fatigue severity subscale of the CIS

409

188

221

6 months

12 months

The intervention was effective in reducing fatigue (B =-4.36, p = 0.020, d = 0.21).

Adjusted: 6 months

p = 0.030

95% CI [− 7.87 to − 0.39] (d = 0.21)

Adjusted: 12 months

p = 1.000

95% CI [− 3.88 to 3.88] (d = 0.04)

Between- group differences at 12 months from baseline on emotional ( p = .611, d = 0.04) were non-significant

The intervention group remained fairly stable in fatigue between 6 and 12 months from baseline, but the control group slightly improved over time, leading to non-significant group differences at 12 months from baseline.

Yun 2017 [42]

EORTC QLQ-C30 fatigue score

174

57

117

12 months

From baseline to 12 months, the LP group, relative to the UC group, showed a significantly greater decrease in the EORTC QLQ-C30 fatigue score (p = 0.065)

3 months: p = 0.214 12 months: p value = 0.010**

Yun 2012 [43]

Brief Fatigue Inventory (BFI) and Fatigue Severity Scale (FSS)

273

136

137

3 months

BFI:

p < 0.01

95% CI − 1.04 to-0.27

Cohen’s d = 0.29

FSS:

p < 0.01

95% CI − 0.78 to − 0.21

Cohen’s d = 0.27

Compared with the control group, the intervention group had an improvement in fatigue as shown by a significantly greater decrease in BFI global score (-0.66 points; 95% CI − 1.04 to − 0.27) and FSS total score (− 0.49; 95% CI − 0.78 to − 0.21).