Many of us are doing ongoing prospective research, yet COVID-19 has paused our work for a while or even months because of the city shut-down. There is an unexpected huge increase in lost to follow-up in our research clinics.
How should we deal with these cases, and the associated data?
The selection bias caused by the lost to follow-up cannot be adjusted by study design, as it is started already. What methods can we use to adjust instead?
Can anyone simply explain how inverse probability-of-censoring weighted estimation technique work on this issue?
How to run it practically, e.g. by SPSS software? Or other higher level of statistical software is needed?
How about stratification-based methods or weighted methods? How are they working actually?
Any practical guide available online?
Great thanks in advance with all your help!
This will ultimately be a problem with "missing data," which is a complex topic. If you are not familiar with the basic issues involved in missing data, the Wikipedia article on this topic is a good introduction.
There are a lot of resources and articles online talking about all kinds of bias, which I know already. But there are minimal resources that talked about the way to tackle it while running the research.
Anyone got any good ideas on how to adjust?
Some papers just try to assume all LTFU lost to follow up as +ve or -ve to test the ultimate results.
And paper point out that the reporting of LTFU is not much in general
Reporting of Lost to Follow-Up and Treatment Discontinuation in Pharmacotherapy and Device Trials in Chronic Heart Failure
A Systematic Review
Article Reporting of Lost to Follow-Up and Treatment Discontinuation...
Also
Vervölgyi, E., Kromp, M., Skipka, G. et al.
Reporting of loss to follow-up information in randomised controlled trials with time-to-event outcomes: a literature survey.
BMC Med Res Methodol 11, 130 (2011).
https://doi.org/10.1186/1471-2288-11-130
There are various techniques, some quite advanced, for dealing with missing data. The first step is to determine if your loss to the sample is "missing at random."
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software