ABSTRACT
As the old adage goes, garbage in, garbage out. Simply put, faulty data compromise studies. Survival estimates are biased if patients are lost to follow up; early results are biased by inclusion of patients who will later be found to be ineligible and by missing information; response estimates are biased if tumor assessments are missed; variability is increased if patients are not treated according to protocol, making it more difficult to detect treatment differences. The following three examples are illustrative.
