Burch 1999.

Methods	Randomised controlled trial Method of randomisation: computer generated Concealment of allocation: sealed envelopes Outcome assessor blinding: blinded research sociologist
Participants	Country: UK Patients referred to day hospital Exclusion criteria: dysphasic, required nursing or occupational therapy > twice per week 163 patients eligible (28 needed day hospital treatment, 21 refused consent, 9 operational problems at day centre) Participants randomised = 105 Baseline function: Median (IQR) Barthel Index 15 (12‐17) and 15 (11‐17) Male: 36% Age: mean (SD) 80.4 (7.6) years
Interventions	Day hospital: care by multidisciplinary rehabilitation team, principally nursing assessment, occupational therapy and physiotherapy. Median number of treatments (interquartile range) 11.5 (5–20.5). Day centre: rehabilitation provided by a physiotherapist and two support workers. Median number of treatments (interquartile range) 10 (5–14).
Outcomes	12 month follow up Death Institutional care Barthel Index Caregiver Strain Index Philadelphia Geriatric Morale scale Costs
Notes	Total of 105 patients of whom 23 had a stroke diagnosis, 14 osteoarthritis, 13 fracture, 9 Parkinsonism Of the 55 patients randomised to day centre attendance, 10 transferred to day hospital
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer generated
Allocation concealment (selection bias)	Low risk	Quote: "Immediately after consent, subjects were randomly allocated to day hospital or day centre by a sequence of labelled tickets in sealed, opaque envelopes securely kept and opened by a senior ward clerk unattached to the trial team....computer generated blocks of 20."
Blinding of participants and personnel (performance bias) All outcomes	High risk	Reported as single blind which appears to have been the assessor, not participants or personnel
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Outcome assessments were undertaken by a blinded research sociologist Quote: "The interviewer correctly identified 38/55 as day hospital and 20/38 as day centre, yielding kappa = 0.22 indicating poor agreement/successful blinding."
Incomplete outcome data (attrition bias) All outcomes	Low risk	Similar losses per group (˜30%), moderately high but similar reasons reported for both groups
Selective reporting (reporting bias)	Unclear risk	We were unable to obtain the pre‐study protocol so cannot determine risk of reporting bias
Other bias	High risk	18% of participants randomised to day centre attendance transferred to the day hospital leading to possibility of contamination (the experiment and control groups becoming mixed)