Veterinarians' brief guide to confidence intervals, standardized effect size, and number needed to treat: understanding the impact of treatments and disease.

Abstract

Veterinarians in all settings strive to practice evidence-based medicine, but current methods of reporting scientific study results create barriers to clinical translation of findings. The-value is the most commonly used measure of statistical significance, but fails to convey information about clinical relevance of findings. In this article, the limitations of the-value as a standalone measure and the importance of reporting additional metrics such as confidence intervals (CI), standardized effect size (ES), and number needed to treat (NNT) in veterinary research are discussed. Confidence intervals and ES can help describe the magnitude of an observed effect, whereas NNT is a practical estimate of the average number of animals that need to receive a treatment for one additional animal to experience 'success'. Using examples of clinical studies on the effect of osteochondrosis on racing performance, silver nanoparticles for equid wound healing, and bedinvetmab for treatment of canine osteoarthritis, this article demonstrates the importance of reporting CIs, ES, and NNT in addition to-values for providing deeper insights on treatment-impact and clinical-decision making. Incorporating the use of these tools in addition to the-value in veterinary manuscripts will aid practitioners in making informed, patient-centered decisions.