What is it about?

* Predictive values of diagnostic tests like MRI can be misleading if not standardized properly. * Predictive values depend heavily on the underlying rate of disease in the tested population. A test can seem to have a high negative predictive value just because the disease is rare, not because the test is accurate. * The authors propose standardizing predictive values by calculating them, assuming a 50% disease rate, allowing fairer comparisons between diagnostic tests. This removes the influence of differing disease rates between studies. * Standardized predictive values would let readers compare diagnostic test performance more easily without knowing the disease rates in each study. * Standardizing predictive values by assuming a common disease rate makes comparing the real-world usefulness of different diagnostic tests across studies easier. This gives a more accurate sense of test accuracy.

Featured Image

Why is it important?

- Predictive values of diagnostic tests are often misinterpreted or misleading when comparing between studies. This can lead to inaccurate conclusions about a test's usefulness. - Standardizing predictive values removes the bias of different underlying disease rates. This allows for fairer comparisons of test accuracy between studies. - Knowing a test's true negative and positive predictive value is crucial for clinical decision-making. An apparently high negative predictive value can be falsely reassuring if it reflects low disease prevalence rather than test accuracy. - Standardized predictive values allow doctors and researchers to evaluate better and choose diagnostic tests based on their intrinsic accuracy rather than the characteristics of the studied population. - More accurate knowledge of test predictive values leads to better clinical diagnoses and treatment decisions, improving patient outcomes. In summary, standardizing predictive values improves the validity of diagnostic test comparisons, removes bias, and promotes more informed clinical decision-making - ultimately leading to better patient care.


This publication makes an important contribution by highlighting an underappreciated statistical issue that can mislead researchers and clinicians when interpreting diagnostic test results. Predictive values are prone to dramatic shifts based on the underlying disease prevalence within a studied population. However, these values are frequently reported without appropriate context, giving a false impression of a test's intrinsic accuracy. I have seen many examples where impressive negative or positive predictive values reflect little more than low or high disease prevalence among the subjects. I aim to encourage researchers to provide more meaningful metrics when reporting diagnostic test performance by proposing a standardized method to calculate predictive values assuming an equal disease prevalence. This will allow for fairer test comparisons and better-informed adoption of new modalities. It is a simple statistical adjustment that could meaningfully improve study quality and clinical decision-making. Publishing this perspective was important to me because misuse of predictive values is commonplace in the diagnostic literature. As a clinician and researcher, I want to enhance the validity and clinical utility of study findings in my field. A minor methodological change, such as standardizing predictive values, can go a long way in achieving that goal. I hope this publication sparks wider adoption of this approach when evaluating diagnostic tests.

Thomas F Heston MD
University of Washington

Read the Original

This page is a summary of: Standardizing predictive values in diagnostic imaging research, Journal of Magnetic Resonance Imaging, January 2011, Wiley, DOI: 10.1002/jmri.22466.
You can read the full text:




The following have contributed to this page