From conference abstract to full paper: differences between data presented in conferences and journals

Background: We studied the type and frequency of differences between data presented in conference abstracts and subsequent published papers in the fields of infectious diseases and microbiology. Methods: We reviewed all abstracts from the first session of 7 of 15 major research categories presented in the 1999 and 2000 Interscience Conference on Antimicrobial Agents and Chemotherapy. For each selected pair of abstract and related published paper, two independent investigators performed a detailed data comparison. Results: From 190 abstracts reviewed, 68 (36%) were subsequently published as full papers by March 2004. Fifty-two pairs referred to the same study population and period. Differences were found in 30 of 51 pairs, which were further analyzed (point estimate=59%, 95% C.I.: 45-73%). The identified differences were related to both the aims and conclusions of the study (3/30), the study conclusions only (2/30), numbers and/or rates of the studied patients (10/30), numbers or rates of microbiological isolates (9/30), MIC values or K, values (5/30), other pharmacological properties of antibiotics (2/30), odds ratio (1/30), and duration of observation (1/30). Some differences were considered major. In bivariable associations, time to publication (from presentation in the conference to publication of the full paper) was associated with identifiable differences between the conference abstract and the full paper (OR=1.76, 95% CI 0.95-3.24/year of delay, P=0.07). Conclusions: It is reassuring that although we identified several reportable differences, only a very small proportion of studies exhibited differences in their aims and/or conclusions. Researchers may benefit from the above findings in improving the accuracy of presented data.