Castaldi 2015.

Methods	Identification of subsequent full‐length publications Searched electronic database Cochrane Central Register of Controlled Trials, MEDLINE, PubMed, and Google/Google Scholar to July 2012 Person completing the search not reported Searched by all authors and keywords Matched abstract to full‐length publication by One author Main outcome Searched web sites of major Italian journals in public health and hygiene
Data	Included 621 abstracts presented at the 2005 to 2007 Italian Society of Hygiene and Public Health meetings Included all 'good quality abstracts' defined as those with a score of 19 or higher or between 16 and 19 with unscored item for "Inherence"
Comparisons	Proportion of abstracts published, overall and by meeting Mean time to publication, overall and by meeting 'Positive' versus not 'positive' Sample size ≤100 versus >100 Oral versus poster presentation Observational design versus experimental design versus 'revision' Academic versus non‐academic affiliation Very high quality versus high quality versus medium quality Male versus female first author
Outcomes	146 of 621 abstracts published 39/174 abstracts presented at the 2005 meeting, 44/187 at the 2006 meeting, and 63/260 at the 2007 meeting published Proportion of abstracts published by time Mean time to publication = 25.1 months (95% CI = 22.1 to 28.1 months), overall Mean time to publication = 27.4 months (95% CI = 20.8 to 34.0 months) for the 2005 meeting Mean time to publication = 30.5 months (95% CI = 24.6 to 36.4 months) for the 2006 meeting Mean time to publication = 20.2 months (95% CI = 16.6 to 23.8 months) for the 2007 meeting Factors related to proportion of abstracts published included 121/450 'positive' (defined as significant results) versus 4/41 not 'positive' abstract results published 47/220 abstracts with a sample size ≤100 versus 99/400 abstracts with a sample size >100 published 54/149 abstracts presented orally versus 92/472 presented as poster published 114/513 abstracts with observational design versus 21/63 abstracts with experimental design versus 11/44 presenting 'revisions' published 120/422 abstracts originating in an academic center versus 26/199 abstracts not abstracts originating in an academic center published 38/109 'very high quality' abstracts versus 78/326 'high quality' abstracts versus 30/186 'medium quality' abstracts published 62/299 abstracts with male first author versus 84/318 with female first author published
Notes	Other non‐clinical specialties ‐public health Reported receiving no funding
Risk of bias
Item	Authors' judgement	Description
Sampling method?	Yes	Included all abstracts of a specific subgroup, so low risk of bias.
Search for publications?	Yes	Searched 3 databases and web sites of journals.
Follow‐up time?	Yes	The meetings had at least 60 months follow‐up.
Matching?	Yes	Matched by 2 different criteria.
Adjustment for confounding?	Yes	Examined association of meeting year, positive results, sample size, type presentation, study design, academic affiliation, abstract quality, and author gender with publication using multivariable logistic regression analysis.