Schenkman 1998.

Methods	Parallel group design. Participants were stratified according to gender and then randomly assigned using computer‐generated assignment. Randomisation schedule kept in office of statistician until participants were assigned. Method of data analysis not described. Treated as outpatients for 30 hours over 10 to 13 weeks. Assessed at baseline and immediately after treatment. Assessors were blinded.
Participants	27 participants in exercise group, 24 participants in control group. 4 dropouts from exercise group, 1 from control group. No baseline characteristics given for dropouts. Participants mean age 70.6 years (exercise), 71.2 years (control); male/female 18/5 (exercise), 16/7 (control); Hoehn and Yahr 2.6 (exercise), 2.7 (control). Inclusion criteria: Parkinson's disease as diagnosed by a neurologist, Hoehn and Yahr stage 2 or 3, functional axial rotation of 120 degrees or less to either side. Exclusion criteria: hospitalised within past 3 months, PD drugs changed in last month, other neurological disorders, Folstein MMSE < 23.
Interventions	Exercise: individual exercises to improve spinal flexibility and coordinated movement. Standardised programme included a series of exercises divided into 7 graduated stages, from supine to standing. Exercises learned at each stage are continued throughout with progressively higher level activities added. Exercises are incorporated into daily routine at end of formal training sessions. Control: no treatment. ('Wait listed' for exercise programme). Drug therapy constant during trial.
Outcomes	Functional axial rotation. Functional reach. Timed tests. Timed walk. Cervical and lumbar range of motion. Walking speed. Participants with fluctuations assessed during 'on' time.
Notes	Abstract, further information obtained from author. All 46 participants completed 30 treatment sessions within their allotted time.
Risk of bias
Bias	Authors' judgement	Support for judgement
Eligibility Criteria	Low risk
Randomisation Method	Low risk	Computer‐generated assignment.
Concealment of Allocation	Low risk	Randomisation schedule kept in office of statistician until participants were assigned.
Similarity at Baseline	Low risk
Withdrawals Described	Low risk	10% withdrawals.
Intention To Treat Analysis	Unclear risk	Method of data analysis not described.
Cointerventions Constant	Low risk	Drug therapy was kept constant during the trial.
Blinded Assessors	Low risk	Assessors were blinded to group allocation.