Author Adequate sequence generation? Allocation concealed? Blinding appropriate? Incomplete outcome data addressed? Free of selective reporting? Free of other bias? Jadad score Baum et al. (2003) [10 ] Yes (computer-generated algorithm) Unclear (assignment by opening sealed envelopes supplied in sequence by the study coordinator) Yes (all tests administered by blinded therapists. MMSE administered by two blinded medical students and research nurses) Unclear (13%) after baseline data missing (death/acute illness) One death during follow-up period Unclear (details of nursing home patients provided but not the assisted living patients) No (selection bias suggested in text) 0 Hruda et al. (2003) [11 ] Unclear (no methodology discussion within the text) Yes (assigned in a lottery format ≈ 1 : 2 ratio (control
; intervention
)) Unclear (no methodology discussion within the text) Unclear (no method discussion within the text) Unclear (no method discussion within the text) Unclear (no method discussion within the text) 2 Van de Winckel et al. (2004) [12 ] No (no sequence generation evident) No (randomisation performed by tossing a coin. It is unclear who or how many times the coin was tossed) No (physical tests performed by Physio who performed the exercise groups) Yes (only one dropout—hip fracture) Yes (no selective reporting evident) Yes 0 Thomas and Hageman (2003) [13 ] No (no randomisation) No (no randomisation/blinding No (no blinding) Unclear (trial in place for 6 weeks and authors report “only 1/3 of subjects completing less than 11 sessions.” It is also not clear what happened—were they discontinued?) Yes (no selective reporting apparent) Yes 2 Nicholson et al. (1997) [14 ] No (randomisation and assignment reported to be “not possible”) Yes (randomisation not possible; research assistant blinded to allocation) No (it is not clear who and how subjects were allocated) Unclear (unclear completion or drop-out levels and “some missing data” are also noted) Unclear (no selective reporting apparent; however, data might be incomplete) Yes (no other bias apparent) 2 Witham et al. (2005) [15 ] Yes (computer-generated numbers) No (primary researcher wrote cards with “Exercise” or “Control” out. The cards were then placed into numbered envelopes (1 upwards) which were opened in sequence as each new patient was randomised) No (primary researcher was able to identify sequence of allocation) Yes (82% attendance rate—primary author contacted and provided details of absences) Yes (no selective reporting evident) Yes 0
