PROGNOSTIC ASSESSMENT FROM STUDIES WITH NONRANDOMIZED TREATMENT ASSIGNMENT

In clinical practice, prognostic factors are primarily used to identif y and select patients with a relatively poor prognosis who may need mo re aggressive treatment in order to improve their prognosis. Preferabl y, assessment of the ability of prognostic factors to distinguish thes e high risk patients from low risk patients should take place in the a bsence of such prognosis-modifying adjuvant therapy. Adjuvant therapy may dilute the effect of prognosticators (covariate-treatment interact ions) and such a situation may go unnoticed in tests for interaction b ecause of their low statistical power. These undetected covariate-trea tment interactions will bias the estimated effects of prognostic facto rs in both studies with randomized and non-randomized treatment assign ment but the bias will usually be greater in studies when treatment wa s not randomized. For the evaluation of prognostic;factors, it is ther efore argued that the study population should be restricted to patient s who do not receive adjuvant therapy. This restriction might result i n a less powerful analysis, but it carries the advantage that undetect able biases are avoided. Non-random treatment assignment will not inva lidate this restriction analysis provided that all the prognostic fact ors on which treatment choice was based, are known and incorporated in to the analysis.