I am involved in designing a diagnostic test accuracy study in which the PI wants to compare weight bearing x-ray with gravity stress x-ray (two types of radiography) in diagnosing whether a weber B ankle fracture requires an operative treatment or non-operative treatment.
The PI believes that clinical decisions based on gravity stress x-ray alone may send more than required for operative treatment which obviously is associated with more post-surgery complications.
There is no current reference standard for this diagnosis, the PI wants to make the weight bearing x-ray the gold standard.
Due to ethical reasons they do not want to randomise patients into gravity stress x-rays.
Due to the unavailability of the gold standard we cannot compare these two diagnostic methods in terms of their sensitivity and specificity. Would you be able to advise me on what is the best way forward in designing this trial.
I appreciate that statistical methods are available to compare diagnostic tests in the absence of a reference level, but which method is more practical?
Thanks and greatly appreciate your input on this matter.
Hello Mendis,
There are statistical methods that help identifying groups based on observed characteristics in a given set of data. Try to look after a latent class model. It is used in scenarios in which someone want to estimate the accuracy of a diagnosis when there is no gold standard. A search on PubMed with the terms "latent class analysis" will return a few thousands of papers, such as:
http://www.ncbi.nlm.nih.gov/pubmed/26631997
http://www.ncbi.nlm.nih.gov/pubmed/26661588
Best;
- Badges
- Science method