Table 4. Reviews reporting exercise intervention subgroup differences.

Study	Sub-group
Marques et al.[47]—CFS	Analysis: secondary-tertiary interventions, interventions delivered by psychologists or psychotherapists and those providing minimal contact were more effective compared to those in primary care, delivered by other healthcare practitioners (e.g. exercise/physical therapist, nurse, physiologist) and providing intensive contact. Including psychological components and allowing activity flexibility reported as more effective, but with high levels of heterogeneity.
Larun et al. [41]—CFS	Analysis: graded aerobic exercise (versus treatment as usual) was more effective than anaerobic exercise (versus relaxation). No differences between diagnostic criteria or type of control group.
Puetz et al. [50]–coronary heart disease	Analysis: interventions in non-controlled studies were more effective than in controlled studies. No differences by cardiac rehab intervention type (multifactorial versus exercise).
Heine et al. [23]–MS	Analysis: interventions comparing to non-exercise controls (versus exercise controls) were more effective. Intervention type had an effect: endurance and mixed training reduced fatigue, whereas muscle power and task-orientated training did not.
Andreasen et al. [40]—MS	Observation: the majority of studies reporting an effect had fatigued population at baseline, whereas the majority not reporting an effect were those where the study population was not. No difference by exercise modalities. Differing duration, frequency and intensity of exercise, and the use of different fatigue scales was inconclusive. Social interaction and motor mechanisms were broached as potential mechanism for exercise reducing fatigue.
Asano et al. [51]–MS	Secondary reporting: effect sizes for studies including fatigued participants were typically positive and significant, whereas for non-fatigued participants they were negative and non-significant.