Volpe 2014b.

Study characteristics
Methods	RCT
Participants	Setting: facility, Italy N = 34 Sample: did not describe the source of patients Age (years): mean (SD) intervention group 68 (7) control group 66 (8) Inclusion criteria: diagnosis of ‘clinically probable’ idiopathic Parkinson’s disease; HY stage 2.5 and 3; ability to walk without any assistance; at least two falls in the last year; Mini‐mental State Examination score ≥ 25; no relevant comorbidity or vestibular/ visual dysfunctions, limiting locomotion or balance; stable dopaminergic therapy in the last four weeks Exclusion criteria: history of deep brain stimulation surgery and other conditions limiting hydrotherapy (for example cardio pulmonary disease). Disease severity at baseline: HY stage 2.5 to 3, UPDRS motor score mean (SD) 40.6 (10.8)
Interventions	Exercise 1. Exercise: hydrotherapy focused on perturbation‐based balance training (60 minutes, 5x/week for 8 weeks) 2. Control: land‐based treatment focused on perturbation‐based balance training (60 minutes, 5x/week for 8 weeks)
Outcomes	1. Rate of falls 2. Number reporting a fall‐related fracture 3. Quality of life (PDQ‐39) Other outcomes reported but not included in this review
Duration of the study	10 weeks
Funding source	No funding
Notes	Fall data collected: Falls which occurred two months prior to the trial and during the 2 month trial period were collected by falls diary or telephone interview
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A random component in the sequence generation was described. Quote: "For the allocation of the participants, a computer‐generated list of binary random numbers was used.”
Allocation concealment (selection bias)	Low risk	Allocation concealment was described as by central allocation. Quote: "The sequence was concealed and the following number (0: Group 1; 1: Group 2) was disclosed by a person not involved in the enrolment process, every time a new patient was added.”
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Participants and intervention (exercise) delivery personnel not blinded to group allocation but impact of non‐blinding unclear.
Blinding of outcome assessment (detection bias) Falls and fallers	Unclear risk	Unclear if personnel collecting fall information blinded to group allocation.
Blinding of outcome assessment (detection bias) Fractures	Unclear risk	Insufficient information to permit judgement. No fractures reported (as no injurious falls) and unclear how data regarding fractures was collected.
Incomplete outcome data (attrition bias) Falls	Low risk	See appendix for method of assessment
Selective reporting (reporting bias)	Unclear risk	Insufficient information to permit judgement as unable to find a published protocol or trial registration.
Method of ascertaining falls (recall bias) Falls and fallers	Unclear risk	The study used some form of concurrent collection of data about falling‐ i.e. falls diaries, but frequency of follow‐up by the researchers was not reported.