Chang 2010.

Study characteristics
Methods	Study design Quasi‐RCT Study dates Duration of follow‐up: 8 weeks Time frame: August to November 2008
Participants	Study characteristics Setting: multicentre (2 HD units in a medical centre in northern Taiwan) Country: Taiwan Inclusion criteria: patients were conscious; able to communicate; on HD for at least 3 months; had Kt/V > 1.1 for the last 3 months; HCT values > 27%; albumin levels > 3.7 g/dL; GOT and GPT values < 50 U/L; able to use a leg ergometer in bed without assistance Exclusion criteria: neuromuscular problems Baseline characteristics Number (analysed/randomised): intervention group (36/44); control group (35/46) Mean age ± SD (years): intervention group (50.8 ± 10.72); control group (52.0 ± 8.7) Sex (M/F): intervention group (26/10); control group (24/11) Dialysis type: HD Mean dialysis vintage ± SD (years): intervention group (6.43 ± 3.91); control group (7.04 ± 4.16) Comorbidities CVD: not reported Diabetes: not reported Hypertension: not reported Depression (clinician diagnosis): not reported
Interventions	Intervention classification Non‐pharmacological intervention Indication: study targeting fatigue Intervention group Intradialytic leg ergometry exercise Control group Sedentary group Co‐interventions Not reported
Outcomes	Outcomes reported Fatigue outcome measures used: validation data available Change in fatigue HD fatigue scale: assessed at baseline, and at 4 and 8 weeks (Appendix 3) Change in physical activity Bouchard’s PAL: assessed at baseline, and at 4 and 8 weeks (Appendix 3) BP: assessed before, during and after exercise Heart rate: assessed before, during and after exercise Peripheral oxygen saturation (SpO2): assessed before, during and after exercise Cardiopulmonary response and signs of physical discomfort (fainting, chest pain or tightness, dyspnoea, nausea, vomiting, muscle or joint pain, or unsteady pedal speed): assessed before, during and after exercise
Notes	Additional information Funding: Taipei Medical University and Shin Kong Memorial Hospital Conflicts of interest/disclosures: none Trial registration identification number: not reported A priori published protocol: not reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "This was a quasi‐experimental clinical trial in a medical centre with two haemodialyses units managed by the same medical and nursing team. The patients were assigned randomly to either unit. The experimental group was recruited from one unit and the control group from another, and participants were pair‐matched based on age and gender." Comment: Since the study was a quasi‐experimental trial, no random element was used in generating the allocation sequence or the sequence was predictable
Allocation concealment (selection bias)	High risk	Method of allocation concealment was not reported in sufficient detail to permit judgement. However, since the study was a quasi‐experimental trial, there was a reason to suspect that the enrolling investigator or the participant had knowledge of the forthcoming allocation. No imbalance between intervention groups was apparent
Blinding of participants and personnel (performance bias) All outcomes	High risk	Not reported. However, interventions were different and participants and/or investigators could be aware of the treatment assigned
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: "Subjects were interviewed by a research assistant to fill‐out the fatigue scale and Bouchard’s PAL on enrolment, during the fourth week and the eighth week of their haemodialysis visits. The research nurse is not a staff working in these haemodialysis units. She collected data independently and did not participate in patient care." Comment: The outcomes were assessed with an appropriate measure, without differences between groups. However, subjective measures were used, it was not stated whether outcomes were assessed without knowledge of treatment allocation, and knowledge of treatment assignment may have influenced reporting. Participant/investigators beliefs about the superiority/inferiority of either intervention could have influenced their assessment of the outcome, but there was no evidence that this was likely. However, objective and subjective outcomes were assessed. It was not stated if the interviewer was blinded to the treatment allocation
Incomplete outcome data (attrition bias) All outcomes	High risk	Quote: "From August to November 2008, there were 44 and 46 subjects in each unit who met the criteria and were invited to participate. Fourteen refused in the beginning as they were unwilling to participate. Five subjects dropped‐out in later stages for various reasons (Figure 1). Thirty‐six subjects (80%) in the experimental group and 35 patients (76%) in the control group completed the study." Comment: 36/44 participants in the intervention group and 35/46 participants in the control group completed the study (> 5% lost to follow‐up with differences between groups). Reasons for discontinuations were not reported
Selective reporting (reporting bias)	High risk	Information about the protocol and the statistical analysis plan were not reported. Fatigue was reported using multiple eligible outcome measurements (scales, time points). Fatigue was reported in a format that was extractable for meta‐analysis. All outcomes that should be addressed (fatigue, cardiovascular disease, and death) were not reported
Other bias	Low risk	There was no evidence of different baseline characteristics, or different non‐randomised co‐interventions between groups. Funding was unlikely to influence the data analysis and authors had no conflicts of interest. The study seemed to be free from other source of bias