Loss to follow up

Loss to follow up is a big problem in the medical literature, fortunately, in the ICU, it is easy to know what happened to patients. Outpatient studies are a lot more likely to have loss to follow up, particularly if they have outcomes that require long follow ups and multiple visits.

There is a rule of thumb that less than 10% loss of follow up is acceptable, however, it depends on the absolute risks.

If you have a difference between the groups of 1% mortality, but your loss to follow up is 10%, then the truth could be that the real difference is 11% more mortality for the same group or that the mortality is as much as 9% worse for the other group (if we take the extremes! - assuming worse-case scenario)

Who gets lost to follow up.... it is usually patients with a different prognosis that the ones that stayed. Maybe they were more non-adherent, maybe they died... we don't know exactly, and we can make mistakes if we assume.

In the literature, we have dealt with the uncertainty in many ways, we have assumed they fared the same, we have carried the last observation forward or we have assumed the worst-case scenario. The only way to know for sure is ensure correct and complete follow up.