A letter to the editor published in the Journal of Magnetic Resonance Imaging examines the usefulness of standardizing predictive values in diagnostic imaging research. The author, Dr. Thomas F. Heston, argues that reporting predictive values adjusted to a standardized disease prevalence of 50% provides more meaningful information to clinicians compared to unadjusted predictive values. He explains that diagnostic tests are most commonly used when the pre-test probability is around 50%, so predictive values tailored to this prevalence are most relevant. Dr. Heston suggests researchers present not only raw predictive values, but also values standardized to disease prevalences of 25%, 50% and 75%. This approach accounts for the variability in disease prevalence between patients and allows clinicians to better judge the utility of a test. The letter emphasizes the importance of presenting research findings in a clinically applicable way.
Citation: Heston, Thomas F. (2014). Standardized Predictive Values. Journal of Magnetic Resonance Imaging, 39(5), 1338. https://doi.org/10.1002/jmri.24564