Survival Analysis-missing time points?

#1
Hi everyone,
Im new to survival analysis and I was hoping I could get some help. I am trying to perform survival analysis (time to event) on a year long trial where participants are sampled at multiple time points and I am looking for time of event (infection occurrence) and also for time of infection recurrence among different treatment groups. During the trial some participants missed a visit in the middle of the trial but showed up for the next visit for example: 1, 2, 3, 4, (missed visit 5), 6, 7, 8, 9, (missed visit 10), 11, 12.
Since the condition I am looking for is transient, it could have occurred in the censored time point (but also at any time point between visits).
What do I do with the missing time points in terms of the statistical analysis. Do I still include data from those subjects in my survival curves, do I censor at the point of missed sample collection even though I have sample information following that visit?

Thank you all in advance!
 

hlsmith

Not a robit
#2
Hmm, I personally have not come across this in my own data. So can a person have an infection and it resolves all within the missing period - so you have no idea? What is the infection to recovery time? Can it happen between collected time periods as well.

My first inkling would be to say, look at the literature to see how others handle this when dealing with this infection and time series data. I am curious to hear your response to my above question before considering suggestions.

thanks.
 
#3
treatment is given over a period of 2-5 days and recovery can be anywhere during that period or within that period. The infection can certainly happen between collected time periods as well however we are likely to see evidence during one of the sampling periods since no treatment is given again until sample collection (and if detected). For example when testing the effect of smoking on the effect of antivirals on influenza clearance and persistence/re-occurrence where subjects are only sampled for viral testing once a month or once every 6 weeks, an infection can certainly happen in between (and may be asymptomatic) and resolve on its own.
 

hlsmith

Not a robit
#4
Well it is what it is, what is the variable of interest and would you imagine it can be confounded (associated) with missing visits. If so you have a problem, if not - and missing visits are indenpendent of everything and ideally balanced between variable of interest groups then you could move forward but list this as a limitation and see if you can come up with any sensitivity analyses.