Bias increasingly affects results as the number of patients lost expands.
Lost patients often have different prognoses than those who remain.

•  Worse prognosis:  patients disappear because of adverse outcome (?death)
•  Better prognosis:  If you feel better why bother with follow-up?

Authors should re-calculate outcomes of patients lost to follow-up as if all in the treatment group did poorly and all patients in control group did well.  If conclusions change then you should be less confident of the validity of the study.