Validity of observational evidence on putative risk and protective factors: appraisal of 3744 meta-analyses on 57 topics

Table 4 Changes in number of associations that are graded as having convincing evidence when one criterion is dropped or replaced by a more lenient version

Credibility assessment	N associations (total=1457)^a	Proportion
Convincing	99	6.8%
Replace P < 10⁻⁶ by <0.001	134	9.2%
Replace P < 10⁻⁶ by < 0.05	142	9.7%
Without the minimum number of cases criterion	149	10.2%
Without the largest study at p < 0.05 criterion	103	7.1%
Without the 95% prediction interval criterion	106	7.3%
Without the heterogeneity I²<50% criterion	159	10.9%
Without the small study effects criterion	122	8.4%
Without the excess significance criterion	111	7.6%

^aThese are the associations that are statistically significant (P < 0.05) and also have information on all criteria

ISSN: 1741-7015