Statistical considerations when assessing outcomes following treatment forprostate cancer

Purpose: We explore the impact of study designs, biases, outcome variables and statistical techniques when interpreting studies concerning prostate ca ncer management. Materials and Methods: Examples from the current literature and a recently assembled population based sample of patients 55 to 75 years old at diagnos is identified by the Connecticut Tumor Registry as having newly diagnosed l ocalized prostate cancer between 1971 and 1984 are provided to assist the r eader to understand the principles discussed. Results: Most reports concerning prostate cancer outcomes suffer fr om obvi ous and subtle biases that confound the reader's understanding of the impac t of different treatment alternatives. Conclusions: By remaining vigilant to these confounding issues, clinicians and patients can gain greater insights into the medical literature and can make individual interpretations concerning the potential impact of treatmen t interventions on men with prostate cancer.