Longitudinal methods for evaluating therapy.

Abstract

Longitudinal studies of therapy are unparalleled in their ability to ensure scientifically cogent results by clearly separating cause (therapy) from effect (therapeutic outcome), by allowing incorporation of duration of follow-up into the assessment of comparative outcomes, and by permitting the use of safeguards against non-concurrent therapies, unequal prognostic susceptibilities, unequal therapeutic performance, and unequal detection of outcomes. The price of these advantages includes the reduced ability of longitudinal studies to evaluate therapies whose outcomes occur rarely or with long latencies, as well as the specific logistic, ethical, and design problems that limit the ability of clinical trials to answer many of the therapeutic questions which are important to clinical practice. The implications of these limitations are that longitudinal evaluations may not be suitable for addressing many therapeutic issues--particularly those involving preventive therapies--and that even when longitudinal studies are suitable, randomized trials may not be logistically feasible or ethically permissible to answer every clinically important question. One of the important challenges for research in this area is to develop alternative methods for assessing therapeutic effectiveness.