In Reply: As noted in our article, an important limitation of nonrandomized studies is the potential for biases associated with unmeasured confounding variables. Traditional statistical methods have been limited in the ability to adjust for unmeasured or unknown biases.
Instrumental variable analytic (IVA) methods, like the one we used, have been used to better overcome these challenges and limitations. Previous studies have shown that IVA methods can very effectively adjust for unmeasured or unknown imbalances,1- 2 such as might have occurred with PSA values. Although we could not unequivocally document balance in PSA values, examination of other factors, such as age, comorbidity, stage, and grade, demonstrated balance in the comparison groups, suggesting that the IVA method accomplished its goal. In addition, the results of our study were consistent with other similar studies,3- 6 further suggesting that the results are likely valid.