Lead time bias vs overdiagnosis

Lead time bias vs overdiagnosis

by Thomas Newman -

Hi, Students!

Question received via email:

Hi Dr. Newman,

I’m having a tough time separating overidagnosis/pseudodisease as a bias in contrast to lead time or length time bias.  Any suggestions on how to frame this in my mind?  It seems like overdiagnosis is a form of lead time bias. 

Thank you,

XXX

ANSWER:

Lead time bias adds years to survival by moving the date of diagnosis (from which survival is counted) sooner.  But the most it can add to survival is the length of the latent phase, I.e. the time period during which the disease can be diagnosed by the test, but is not yet symptomatic.  You diagnose the disease earlier, but it still a disease.  

Overdiagnosis is sort of like lead time bias with a latent phase equal to the patient's life expectancy.  That is, the disease would NEVER have become symptomatic if it had not been diagnosed by screening.   The entity you are diagnosing is thus sometimes called pseudodisease.

Does this help?

Tom

Re: Lead time bias vs overdiagnosis

by Christine -

Hi Tom,

Yes it does.  Lead time and length time are now seeming to be very similar.  Lead time bias seems to refer to a homogenous disease versus length which more applies to heterogenous disease.  But given that most (all) disease are heterogenous, don't you have lead time bias in length time bias?
 
I should have asked more questions in class last week—if you don’t have a chance to clarify lead and length time via email, I’ll ask you in class tomorrow.  
 
Thank you again!
Chris

Re: Lead time bias vs overdiagnosis

by Thomas Newman -

Yes, it's true that when you have length bias, you almost always have lead time bias, because lead time bias occurs any time you count survival from diagnosis and you make the diagnosis earlier, which happens with screening.  But, as you say, you could have lead time bias without length bias if the disease had a uniform natural history.  Maybe pancreatic cancer or some other one that is pretty much always bad pretty quick...

Tom