Nader 2009.

Methods	Identification of subsequent full‐length publications Searched electronic database MEDLINE to January 2007 Person completing the search not reported Searched by all authors and keywords Matching criteria not reported Contacted abstract authors directly
Data	Included 125 abstracts presented at the 1999 to 2001 International Geographic Medicine Congress meetings Included abstracts where author responded to inquiry (125 of 250 randomly selected abstracts)
Comparisons	Proportion of abstracts published 'Positive' versus not 'positive' Sample size less than or equal to 100 versus between 101 and 500 versus > 500 Oral versus poster presentation 'Experimental' design versus 'observational' design Multi‐centered versus single center 'Specified' sponsor versus no 'specified' sponsor
Outcomes	34 of 125 abstracts published Proportion of abstracts published by time not reported Factors related to proportion of abstracts published included 34/96 'positive' (defined as experimental better than control) versus 1/26 not 'positive' abstract results published 23/81 abstracts with sample size of less than or equal to 100 versus 8/33 abstracts with sample size between 101 and 500 versus 3/11 abstracts with sample size > 500 published 14/30 abstracts presented orally versus 20/95 abstracts presented as posters published 19/47 abstracts with 'experimental' design versus 16/78 abstracts with 'observational' design published 10/18 abstracts with multiple centers versus 27/107 abstracts with a single center published 25/46 abstracts 'specifying' a financial sponsor versus 10/79 abstracts not 'specifying' a financial sponsor published
Notes	Other clinical specialties ‐ geographic medicine Funding not reported
Risk of bias
Item	Authors' judgement	Description
Sampling method?	No	Included abstracts only if author responded to inquiry; response rate was 50% (125/250).
Search for publications?	Yes	Searched 1 database and contacted abstract authors.
Follow‐up time?	Yes	All meetings had at least 6 years follow‐up.
Matching?	Yes	Matched by author contact.
Adjustment for confounding?	Yes	Examined association of positive results, type of presentation, study design, multi‐center status, and funding status with publication using multivariable logistic regression analyses.