Data extraction
The first 20 search results were independently screened by G.L. and L.K. to assess exclusion procedure reliability. No paper was excluded incorrectly. All other papers were screened by L.K. and E.L.H. If exclusion was unclear, L.K., E.L.H. and G.L. discussed and reached consensus. Data extracted from the papers (by L.K. and E.L.H.) included methodological characteristics; description of the intervention; whether the intervention was applied to the person with dementia, family caregivers or staff; statistical methods; length of follow-up; diagnostic methods; and summary outcome data (immediate and longer-term). Paper quality, including bias, was scored independently by L.K. and E.L.H., discussing discrepancies with G.L. and/or G.B. They used Centre for Evidence-Based Medicine (CEBM) RCT evaluation criteria (http://www.cebm.net/index.aspx?o=1025); this approach gives points for randomisation and its adequacy, participant and rater masking, outcome measures validity and reliability, power calculations and achievement, follow-up adequacy, accounting for participants, and whether analyses were intention to treat and appropriate. Possible scores range from 0 to 14 (highest quality). Where a randomised design was used but the intervention was not compared with the control group, we considered this a within-subject design, for example the study by Raglio et al.19 We assigned CEBM evidence levels as follows:
1.level 1b: high-quality RCTs (these were at least single-blind, had follow-up rates of at least 80%, were sufficiently powered, used intention-to-treat analysis, had valid outcome measures and findings reported with relatively narrow confidence intervals);
2.level 2b: lower-quality RCTs.
Data extraction
The first 20 search results were independently screened by G.L. and L.K. to assess exclusion procedure reliability. No paper was excluded incorrectly. All other papers were screened by L.K. and E.L.H. If exclusion was unclear, L.K., E.L.H. and G.L. discussed and reached consensus. Data extracted from the papers (by L.K. and E.L.H.) included methodological characteristics; description of the intervention; whether the intervention was applied to the person with dementia, family caregivers or staff; statistical methods; length of follow-up; diagnostic methods; and summary outcome data (immediate and longer-term). Paper quality, including bias, was scored independently by L.K. and E.L.H., discussing discrepancies with G.L. and/or G.B. They used Centre for Evidence-Based Medicine (CEBM) RCT evaluation criteria (http://www.cebm.net/index.aspx?o=1025); this approach gives points for randomisation and its adequacy, participant and rater masking, outcome measures validity and reliability, power calculations and achievement, follow-up adequacy, accounting for participants, and whether analyses were intention to treat and appropriate. Possible scores range from 0 to 14 (highest quality). Where a randomised design was used but the intervention was not compared with the control group, we considered this a within-subject design, for example the study by Raglio et al.19 We assigned CEBM evidence levels as follows:
1.level 1b: high-quality RCTs (these were at least single-blind, had follow-up rates of at least 80%, were sufficiently powered, used intention-to-treat analysis, had valid outcome measures and findings reported with relatively narrow confidence intervals);
2.level 2b: lower-quality RCTs.
การแปล กรุณารอสักครู่..
