7. Summary of other systematic reviews with meta‐analysis to address time spent in rehabilitation.

Review	Type of rehabilitation	Key findings (in relation to time spent in rehabilitation)	Agreement/disagreement with this review
Langhorne 1996	Physiotherapy	There was a non‐significant reduction in the chance of death. The pooled measures of impairment and disability found no significant results.	This review found no difference in the risk of SAE/death with additional therapy. There is limited comparison with the pooled measures.
Kwakkel 1997	Rehabilitation	Small effect in favour of additional treatment seen for ADLs. Effect seen for functional outcomes in favour of additional treatment. No effect seen for neuromuscular outcomes; however, following post hoc analysis to control for organisational setting and blinding, there was an effect.	This review found no effect for ADLs and no effect for activity measures of the ULs and LLs. This disagrees with Kwakkel 1997.
Kwakkel 2004	Exercise therapy	Small effect found for ADL and walking speed. No effect seen for UL outcomes (measured with the Action Research Arm Test). For the ADL outcomes, a cumulative meta‐analysis was undertaken. This found that at least an additional 16 hours of exercise therapy is required to elicit a 4–5% change in outcome measure.	This review found no effect for ADL and LL activity measures (such as walking), which disagrees with Kwakkel 2004 This review found no effect for activity measures of the UL (such as the Action Research Arm Test), in agreement with Kwakkel 2004
Galvin 2008	Exercise therapy	No effect found for UL measures (pooled functional and impairment measures). No effect found for LL measures (pooled functional and impairment measures). Effect seen in favour of additional therapy for ADL measures (as measured using the Barthel Index).	This review split functional and impairment measures of the UL and LL, so this outcome is not comparable. This review saw no effect for ADL measures, in contrast with the Galvin 2008 review.
Cooke 2010a	Exercise‐based therapy	Meta‐analysis was undertaken for hand grip force/strength at end of treatment. This favoured the control treatment. For Motricity Arm measured at first follow‐up, there was an effect in favour of experimental treatment. There was no effect for measures of UL function (Action Research Arm Test). Comfortable walking speed showed an effect in favour of control treatment at first time point, but a non‐significant finding at second time point. Rivermead mobility showed a non‐significant effect.	This review found an effect for motor impairment of the UL, which is in contrast with some findings of Cooke 2010 (which split measures of motor impairment of the UL). This review found no effect for measures of UL function, in agreement with Cooke 2010. This review found no effect for activity measures of the LL, which is in contrast to the findings of Cooke 2010 at the first time point.
Veerbeek 2011	Lower‐limb exercise therapy	Beneficial effect of more therapy seen for walking ability, comfortable walking speed, and maximum walking speed. No effect seen for basic ADLs, but an effect seen for extended ADLs.	This review did not find an effect for LL activity or ADLs, as Veerbeek 2011 did.
Lohse 2014	Therapy	There was an overall beneficial effect of receiving more therapy than receiving less (all outcomes combined). A meta‐regression was performed using 4 different models, which controlled for the linear and non‐linear effects of time and time since stroke. They concluded that there was a significant, positive relationship between amount of time scheduled for therapy and improvement on outcome measures. This relationship was not affected by time since stroke, but there was a potentially non‐linear effect of time.	There is limited comparison with this review, as Lohse 2014 combined outcomes. The findings of the scatter diagrams in this review agree with the findings of Lohse 2014.
Pollock 2014a	Physical rehabilitation	Subgroup analyses found that there was a greater effect size in studies with a greater amount of time spent in rehabilitation, with an indication that 30–60 minutes once per day for 5–7 days per week was beneficial, but that more than once‐daily intervention may provide even greater benefit.	This study found no evidence that supported a specific therapy schedule.
Sehatzadeh 2015	Physiotherapy	Greater amount of therapy led to greater improvements in UL activity. No significant difference in measures of mobility with increased amount of therapy. No significant difference in ADL.	This review found no effect for UL activity with more rehabilitation, as Sehatzadeh 2015 did. This review agrees with Sehatzadeh 2015 about lack of effect for mobility (LL activity) and ADL.
Schneider 2016	Rehabilitation	Found that additional therapy had a beneficial effect on UL and LL activity immediately after training. Subgroup analysis showed that there was a greater effect in studies that provided a large increase in therapy, compared to a small increase.	This review found no beneficial effect on UL and LL activity, as Schneider 2016 did. This review found a greater effect when there was a larger difference in amount of rehabilitation between study groups, which agrees with Schneider 2016.
French 2016	Repetitive task training	There was no difference between subgroups for trials that delivered 0–20 hours of therapy or ≥ 20 hours of therapy for UL function or LL function.	This trial found no effect for additional time spent in rehabilitation for activity measures of the UL and LL, which agrees with French 2016.

ADL: activities of daily living; LL: lower limb; SAE: serious adverse event; UL: upper limb.