ACCOUNTING FOR TIME-DEPENDENT COVARIATES WHOSE LEVELS ARE INFLUENCED BY EXPOSURE STATUS

When measuring the association between an exposure and disease, one mu st decide whether to account for confounding or modifying variables wh ose levels are altered by the presence of the exposure. For example, t o assess the impact of cessation of unopposed estrogen therapy on the occurrence of endometrial cancer, a researcher needs to consider the d uration of the estrogen therapy, a strong risk factor for endometrial cancer, as a potential confounder or effect modifier. Duration of estr ogen therapy, however, is itself influenced by the decision to stop th e therapy (the ''exposure'' of interest). In such a case, two distinct approaches may be taken, depending upon the question being considered . One may wish to assess the degree to which the exposure predicts dis ease incidence, over and above the additional variable, at some later point in time. In this case, it is appropriate re, consider the value of the other variable (for example, duration) at that later time. On t he other hand, one may also wish to measure the rate of disease beginn ing at the time of cessation of the exposure, relative to the correspo nding rate in persons with continuing exposure. Here, the most appropr iate analysis considers the level of the confounding variable (for exa mple, duration) measured only until the time the exposure of interest occurs (for example, cessation of unopposed estrogen therapy). Example s are given to illustrate that the specific question being addressed d ictates the handling of covariates of this type.