Dear All,
I have a question that I could not find its answer over the last two days. I have time-to-event data, with outcome miscalssification and known sensitivity and specificity. The outcome was derived from administrative databases using a specific algorithm. However, the algorithm is not accurate. The algorithm may misclassify individuals who had the event as event-free and that did not experience the event as diseased. The probability of this error (the sensitivity and specificity of the algorithm) is known.
My question is related to accounting for outcome misclassification and incorporating the sensitivity and the specificity in the simulation for Cox model (proc phreg). I have seen some people using multiple imputation with internal validation data, but I do not have access to the validation data (data used to calculate the sensitivity and specificity). Alternatively, I have seen other people used Monte Carlo simulation to generate confidence band for the hazard ratios, but I am not sure how I can incorporate the sensitivity and specificity in the simulation (actually, I am not sure I need to incorporate the sensitivity and specificity in the simulation).
My questions:
1/Do I need to incorporate the sensitivity and specificity in the simulation? And is it possible?
2/ Is there any other approach to deal with the outcome miscalssification in survival analysis?
3/ Is there any SAS macro or documetation that I can refer to?
Thanks in advance for your help and support,
Musfer
I don’t know of any statistical methodology that deals with this, so in those cases my go to is simulation.
If you know the errors though (but not actuals) I would simulate the data 10000 times and get my CI and estimates that way.
If you’re not familiar with bootstrapping or simulation in SAS, read the paper don’t be loopy by David Cassell.
@Musfer wrote:
Dear All,
I have a question that I could not find its answer over the last two days. I have time-to-event data, with outcome miscalssification and known sensitivity and specificity. The outcome was derived from administrative databases using a specific algorithm. However, the algorithm is not accurate. The algorithm may misclassify individuals who had the event as event-free and that did not experience the event as diseased. The probability of this error (the sensitivity and specificity of the algorithm) is known.
My question is related to accounting for outcome misclassification and incorporating the sensitivity and the specificity in the simulation for Cox model (proc phreg). I have seen some people using multiple imputation with internal validation data, but I do not have access to the validation data (data used to calculate the sensitivity and specificity). Alternatively, I have seen other people used Monte Carlo simulation to generate confidence band for the hazard ratios, but I am not sure how I can incorporate the sensitivity and specificity in the simulation (actually, I am not sure I need to incorporate the sensitivity and specificity in the simulation).
My questions:
1/Do I need to incorporate the sensitivity and specificity in the simulation? And is it possible?
2/ Is there any other approach to deal with the outcome miscalssification in survival analysis?
3/ Is there any SAS macro or documetation that I can refer to?
Thanks in advance for your help and support,
Musfer
Are you ready for the spotlight? We're accepting content ideas for SAS Innovate 2025 to be held May 6-9 in Orlando, FL. The call is open until September 16. Read more here about why you should contribute and what is in it for you!
ANOVA, or Analysis Of Variance, is used to compare the averages or means of two or more populations to better understand how they differ. Watch this tutorial for more.
Find more tutorials on the SAS Users YouTube channel.