Table 2. Methodological Quality Assessment.
| Quality criteria | Inception cohort | Complete follow up of all trials | Publication ascertained through personal contact with investigators | Definition of positive and negative findings clearly defined | Comparison of protocol to publication |
| Easterbrook, 1991 [24] | Y | N (25% lost to follow up) | Y | Y (positive: p<0.05/striking, negative: p≥0.05/definite but not striking, null: no difference observed between the groups/null findings.) | NA |
| Dickersin, 1992 [23] | Y | Y | Y | Y (positive: p<0.05/statistically significant, negative: suggestive trend but not statistically significant, null: no trend or difference. In terms of importance when statistical tests were not performed: great, moderate or little.) | NA |
| Dickersin, 1993 [3] | Y | N (14% refused to participate) | Y | Y (positive: p<0.05 significant/of great importance, negative: showing a trend in either direction but not statistically significant/moderate importance/no difference/little importance.), | NA |
| Stern, 1997 [4] | Y | N (only 70% of questionnaires were completed) | Y | Y (positive: p<0.05 significant/striking/important/definite, negative: non-significant trend 0.05≤p<0.10 or non-significant or null p≥0.10/unimportant and negative) | NA |
| Cooper, 1997 [19] | Y | Y | Y | N (significant and non-significant) | NA |
| Wormald, 1997 [30] | Y | Y | Y | Y (positive: p<0.05, negative: p≥0.05) | NA |
| Ioannidis, 1998 [5] | Y | Y | Y | Y (positive: p<0.05 significant and in favour of experimental therapy arm or any arm when there is no distinct control, negative: nonstatistically significant findings or favouring the control arm ) | NA |
| Pich, 2003 [27] | Y | Y | Y | NA | NA |
| Cronin, 2004 [20] | Y | Y | Y | U (study showed effect) | NA |
| Decullier, 2005 [7] | Y | N (only 69% of questionnaires were completed) | Y | Y (confirmatory/inconclusive/invalidating) | NA |
| Decullier, 2006 [22] | Y | N (only 80% of questionnaires were completed) | Y | Y (scale from 1 to 10 for not important to very important) | NA |
| Hahn, 2002 [26] | Y | Y | Y | NA | Y |
| Chan, 2004a [18] | Y | Y | Y | Y (positive: p<0.05, negative: p≥0.05) | Y |
| Chan, 2004b [17] | Y | Y | Y | Y (positive: p<0.05, negative: p≥0.05) | Y |
| Ghersi, 2006 [25] | Y | Y | Y | Y (positive: p≤0.05, negative: p>0.05) | Y |
| Von Elm, 2008 [29] | Y | Y (for drug trials) | Y | Y (positive: p<0.05) | Y |
| Turer 2007 [28] | Y | Y (unable to contact trialists in 9%) | Y | NA | NA |
| De jong 2010 [21] | Y | N (20% did not respond to questionnaire) | Y | NA | NA |
| Blumle 2008 [31] | Y | Y (87% response overall, unclear on RCTs) | Y | NA | Y but data not yet published |
| Hall 2007 [32] | Y | Y | Y | Y statistically significant findings (i.e., major outcome variable under study reported to be statistically different from comparator at P<0.05 level) | NA |
Y yes.
N no.
U unclear.
NA Not applicable.