Hi,
I'm analyzing data from an oncology trial and trying to perform statistical analysis including cox regression analysis and paired t-tests.
My data includes laboratory test values, including tumor markers.
In a few cases, tumor marker values were higher than upper limit of quantification, which returns results displayed in forms of ">10000"
How should I impute the data for quantification? For values of lower than detectable limit I've seen imputations using half of lower limit of detection. Is there a similar rule of thumb for data upper than quantifiable limit?
Thanks in advance.
One way is to increase the upper threshold of the device and the method you are using by calibrating the measuring device with higher concentrations of standard(s). If that is not achievable depending on the type of system you are working with, another way is to dilute the samples to fall within quantifiable range. Calculate the dilution factor and use it to normalize or scale the readings. Moreover, if the two options are not feasible, think of using another method of quantification and if possible orthogonal means of verifying your measurement. I hope this helps.
Best,
Ade.
- Badges
- Science topic
- Similar topics
- Medicine
- Public Health