Skip to main content

Table 4 Post-hoc comparison of abstracts that were subsequently published in lower versus higher impact journals

From: Publication bias in animal research presented at the 2008 Society of Critical Care Medicine Conference

Potential predictor variable Published in lower impact (n = 32) Published in higher impact (n = 30) p valuea
Type of presentation
 Oral (vs. poster) presentation 7 (22%) 7 (23%) 0.99
 Research location in North America 17 (53%) 23 (77%) 0.07
Methodological quality variables
 Randomized 7 (22%) 10 (33%) 0.40
 Blinded 0 (0%) 0 (0%)
 Primary outcome given 6 (19%) 7 (23%) 0.76
 Numbers with denominators 1 (3%) 3 (10%) 0.35
 Number of animals stated 14 (44%) 16 (53%) 0.61
 Sample size calculation 0 (0%) 0 (0%)
Ethical quality variables
 Highest species rodent 20 (63%) 15 (50%) 0.32
 >19 animals used 6 (19%) 9 (30%) 0.30
Outcome variables
 Main outcomes positive 28 (88%) 29 (97%) 0.36
 Statistically significant result 17 (53%) 18 (60%) 0.59
Type of animal model
 Sepsis 9 (28%) 18 (60%) 0.02
 Drug used 14 (44%) 10 (33%) 0.44
 Surgery performed 12 (38%) 13 (43%) 0.80
 Animals stated to be killed 19 (59%) 15 (50%) 0.61
  1. aComparisons made using Fisher’s Exact or Chi square test. For published articles, the journal impact factors were: mean 5.2 (SD 4.9), median 4.5 [IQR 2.4, 7.4], range 0.02–30.36; 5 articles were published in journals with impact factor >10