Table I.

Summary of findings based on outcome domains and study design

Certainty assessment
Nr studies	Study design	Risk of bias	Inconsistency	Indirectness		Imprecision	Other considerations	Impact	Certainty
Activity and Participation (follow up: range 3 weeks to 15 months)
2/6a	Randomised trials and observational study designs	not serious	not serious		not serious	not serious	all plausible residual confounding would suggest spurious effect, while no effect was observedi	RCTs: Koopman et al. (3) and Murray et al. (4): Consistency in methods, interventions and outcomes (e.g. 6MWT, 2MWT, TUG), CIs considered not wide, homogenous. Observational: Bertelsen et al. (28), Brogardh et al. (46), Da Silva et al. (37), Davidson et al. (29), Sharma et al. (32), Skough et al. (38), Willen et al. (8): Some consistency in methods and activity outcomes. Interventions varied: strengthening, CV fitness, mixed, CIs considered not wide, homogenous.	RCT only: ⊗⊗⊗⊗ HIGH All study designs: ⊗◯◯◯ VERY LOW
Body function mental and sensory (follow up: range 4 weeks to 10 months)
3/5	randomised trials and observational study designs	seriousc very seriousb,c	not serious seriousd		not serious seriousg	not serious serioush	all plausible residual confounding would suggest spurious effect, while no effect was observedi	RCTs: Koopman et al. (3), Oncu et al. (48) and Murray et al. (4) some consistency in methods, interventions and outcomes (e.g. all fatigue), CIs considered not wide Observational: Bertelsen et al. (28), Da Silva et al. (37), Davidson et al. (29), Dean et al. (25), Sharma et al. (32): Varied effect, CIs not wide, apparent improvement, interventions similar but outcomes varied.	RCT only: ⊗⊗⊗◯ MODERATE All study designs: ⊗◯◯◯ VERY LOW
Body function lower (follow up: range 16 weeks to 10 months)
2/9	randomised trials and observational study designs	very seriousb,c	seriousd		not serious	serioush	publication bias strongly suspected, all plausible residual confounding would suggest spurious effect, while no effect was observedi	Jones et al. (47), Koopman et al. (3): Varied effect, wide CIs, apparent improvement, similar study characteristics (methods, interventions and outcomes).	All study designs: ⊗◯◯◯ VERY LOW
Body function cardiovascular (follow up: range 8 weeks to 10 months)
4/2	randomised trials and observational study designs	very seriousb,c	seriousd		seriousf,g	serioush	all plausible residual confounding would suggest spurious effect, while no effect was observedi	Jones et al. (47), Kriz et al. (26), Koopman et al. (3), Oncu et al. (48): Varied effect, wide CIs, apparent improvement.	All study designs: ⊗◯◯◯ VERY LOW
Body function non-lower (follow up: range 4 weeks to 16 weeks)
3/2	randomised trials and observational study design	very seriousb,c,d	seriouse		seriousf,g	serioush	all plausible residual confounding would suggest spurious effect, while no effect was observedi	Murray et al. (4), Chan et al. (33), Kriz et al. (26): Varied effect, wide CIs, apparent improvement.	All study designs: ⊗◯◯◯ VERY LOW

^aindicates ratio of RCT to non-RCT studies. CI: Confidence interval, CV: Cardiovascular fitness, RCT: randomised controlled trials, TUG: Timed up and go test, 6MWT: Six min walk test, 2MWT: Two min walk test Explanations:

^bno measure of random variability

^climited adjustment of confounding

^dlimited loss to followup, no intention to treat analysis

^esubstantial heterogeneity

^fDifferences in diagnostic criteria

^gDifferences in outcome measures

^hWide Confidence Intervals

ⁱas identified via funnel plot