CAN WE TRUST THE AVIAN MEDICAL LITERATURE: SURVEY AND CRITICAL APPRAISAL OF THE USE OF STATISTICS IN AVIAN MEDICINE FROM 2007 TO 2011

摘要

Statistics are an indispensable tool in health sciences. Medical research is not usually performed by professional researchers but instead by practitioners with limited statistical knowledge. The problem with the misuse of statistics and poor statistical reporting is not well known in avian and zoological medicine. Avian medicine, although evolving in an exponential manner, is still marginal when compared with other clinical fields and may suffer even more from the lack of quality control regarding biostatistical methodologies. All avian research articles that were published in 8 veterinary journals from 2007 to 2011 were reviewed for their use of statistics. Critical appraisal was conducted using standardized statistical guidelines and predetermined checklists. A survey of the type of statistics employed, their complexity, and the prevalence of specific statistical mistakes was conducted. Association between the quality of statistical use and different variables was also investigated by a logistic model. Overall, 35.3% (128/363) of articles were considered statistically unacceptable. Common mistakes pertained to randomization procedures, validation of test assumptions, and underuse of statistical tools. There was a mild positive effect of the presence of a PhD degree holder or a specialty board diplomate and the journal impact factor on the quality of statistics. The results of this study suggest that more stringent quality control measures on biostatistics should be implemented for avian medical articles. Improving the statistical quality of published avian research articles not only enhances the ethics of-research involving live birds but also benefits the overall medical care provided to birds. Copyright 2015 Elsevier Inc. All rights reserved.