Publication bias in clinical trials in cataract therapies: implications for evidence-based decision-making.

Abstract

Evidence-based decision-making is generally based on published evidence. Therefore, if the published evidence is biased, so will the decision-making. One possible bias is the "positive-results" publication bias. This study attempts to characterize this phenomenon in cataract therapy trials. Studies were categorized as "positive" if their results were congruent with the hypothesis and "negative" if not. Secondary outcomes included the influence of funding source and differences in publication metrics between "positive" and "negative" publications. The U.S. National Library of Medicine Clinical Trials database was reviewed for cataract trials, yielding 248 trials. Trials with less than 2 treatment arms, less than 5 participants, or insufficient reporting were excluded. Data were collected on intervention, treatment arms, funding type, publication rates, citation rate, and the impact factor/H-index of journals. Of the 132 trials included, there were 69 positive and 63 negative results. Publication rate for positive results (71%) was significantly greater than negative results (17%) ( P < .01), with no significant difference in the other publication metrics. In conclusion, "negative" result trials are published less frequently, but are equally valued, if published. There are implications for evidence-based medicine with these findings.