Association between characteristics of the published systematic reviews and assessment of publication bias
| Variable | Total (n = 225) | Assessment of publication bias n (%) | RR (95% CI) |
|---|---|---|---|
| Year of publication | |||
| 2002 | 63 | 12 (19.0) | 5.52 (1.63 to 18.8) |
| 2000 | 75 | 10 (13.3) | 3.87 (1.10 to 13.5) |
| ⩽1998 | 87 | 3 (3.4) | 1 (ref) |
| Country of the first author | |||
| USA/Canada | 85 | 11 (12.9) | 1 (ref) |
| United Kingdom | 53 | 6 (11.3) | 0.87 (0.34 to 2.23) |
| Other countries of the EU | 63 | 6 (8.7) | 0.67 (0.26 to 1.72) |
| Other countries | 18 | 2 (11.1) | 0.86 (0.21 to 3.55) |
| Type of journal of publication | |||
| Cardiology | 66 | 9 (13.6) | 1 (ref) |
| Medicine in general | 52 | 9 (17.3) | 1.17 (0.54 to 2.97) |
| Epidemiology/public health | 12 | 4 (33.3) | 2.44 (0.90 to 6.67) |
| Other specialties | 43 | 2 (4.7) | 0.34 (0.08 to 1.50) |
| Cochrane Library | 52 | 1 (1.9) | 0.14 (0.02 to 1.08) |
| Impact factor | |||
| <British Medical Journal | 54 | 7 (13.0) | 1.23 (0.54 to 2.79) |
| ≥British Medical Journal | 171 | 18 (10.5) | 1 (ref) |
| Epidemiologist among the authors | |||
| Yes | 83 | 9 (10.8) | 0.96 (0.45 to 2.08) |
| No | 142 | 16 (11.3) | 1 (ref) |
| Statistician among the authors | |||
| Yes | 30 | 5 (16.7) | 1.63 (0.66 to 4.05) |
| No | 195 | 20 (10.3) | 1 (ref) |
| Number of primary studies included in meta-analysis | |||
| ⩾8 | 115 | 21 (18.3) | 5.02 (1.78 to 14.2) |
| <8 | 110 | 4 (3.6) | 1 (ref ) |
| Type of studies included in meta-analysis | |||
| Observational | 60 | 13 (21.7) | 2.98 (1.44 to 6.16) |
| Experimental | 165 | 12 (7.3) | 1(ref) |