De Bellefeuille 1992.

Methods	Identification of subsequent full‐length publications Searched electronic database Cancerlink; date of search not reported Search completed by investigator Search criteria not reported Matched abstract to full‐length publication by All authors Keywords Contacted abstract authors directly if no publication found
Data	Included 197 abstracts, which was a random sample of 1058 abstracts submitted to the 1984 American Society of Clinical Oncology meeting
Comparisons	Proportion of abstracts published Mean time to publication Cumulative proportion of abstracts published 'Positive' versus not 'positive' Accepted versus rejected for conference presentation RCT design versus non‐RCT design
Outcomes	115 of 197 abstracts published Proportion of abstracts published by time Mean time to publication = 22.8 months Cumulative proportion of abstracts published at 66 months showed proportion published = 57.8% (114/197 abstracts) Factors related to proportion of abstracts published included 48/65 'positive' (defined as significant results) versus 67/132 not 'positive' abstract results published 63/81 abstracts accepted for presentation versus 52/116 rejected abstracts published 20/31 abstracts with RCT design versus 83/166 abstracts with non‐RCT design published
Notes	Oncology Funding not reported
Risk of bias
Item	Authors' judgement	Description
Sampling method?	Yes	Included a random sample of all submitted abstracts.
Search for publications?	Yes	Searched 1 database and contacted authors.
Follow‐up time?	Unclear	Date of search not reported.
Matching?	Yes	Matched by 2 different criteria.
Adjustment for confounding?	Yes	Examined association of positive results, acceptance for presentation, and study design with publication using multivariable logistic regression.