Advertisement
Table 3 Risk of bias assessment for included studies
| Study (Year) | Randomisation method | Allocation concealment | Blinding | Participation at 12 months | Outcome assessment validity reported | Intention to treat (ITT) analysis |
|---|---|---|---|---|---|---|
| Takahashi (2006) [27] | Random numbers generated in Excel | Not stated | Partial. Nurse assessment was blinded | 448/550 (81%) | Yes | No |
| Coates (1999) [28] | Block randomisation | Not stated | Not stated | 1,141/2,208 (52%) | Yes | No |
| Roderick (1997) [29] | Pairs matched by region | Not stated | Not stated | Intervention 86%; control 74% | Yes | Yes |
| Sacerdote (2006) [30] | Random numbers generated by computer | Yes | Outcome assessors and participants stated to be blinded | 2,977/3,179 (93%) | Yes | Yes |
| Stevens (2003) [31] | Not stated | Not stated | Partial. Clinic staff conducting data collection were blinded | Intervention 89%; control 85% | Yes | No |
| Kristal (2000) [32] | Stratified by sex and age | Not stated | Not stated | 1,205/1,459 (83%) | Partial | No |
| Baron (1990) [33] | Not stated | Not stated | Not stated | 329/368 (89%) | Not stated | No |
| Fries (2005) [34] | Not stated | Not stated | Not stated | 516/754 (68%) | Yes | No |
| Beresford (1997) [35] | Table of random numbers | Recruiters and potential participants blind to group allocation | No | 1,818/2,121 (86%) | Yes | No |
| Gann (2003) [36] | Table of random numbers | Not stated | Not stated | 177/213 (83%) | Yes | Yes |
