OPTIMIZING BREAST CANCER MAMMOGRAPHY SCREENING POLICIES CONSIDERING WOMEN'S ADHERENCE BEHAVIORS

Purpose:

Longitudinal studies of several data registries such as the Breast Cancer Surveillance Consortium (BCSC) have shown that women’s adherence to screening mammography guidelines are relatively low. However, currently none of the existing screening mammography guidelines take women’s adherence behavior into consideration. This study proposes a framework to tailor screening recommendations for women based on their adherence behaviors in order to minimize the risk that the breast cancer becomes symptomatic.

Method:

A nonlinear programming model is developed to identify the optimal mammography screening policies while considering uncertainty in women’s adherence behaviors. The objective is to minimize the risk of breast cancer reaching to an advanced stage before it is detected during a screening mammography. Different sources of variation, such as tumor onset time, sojourn time and women adherence behaviors, are incorporated into the model. The adherence behavior is defined by the amount of time a woman delays undergoing a prescribed mammogram. Given a woman’s adherence behavior, the optimal screening policy is obtained under the following constraints: 1) the beginning and ending ages of screening, and 2) the minimum length of interval between two subsequent tests the woman undergoes.

Since the nonlinear model is highly complex and difficult to solve, a heuristic method is applied to extract the near-optimal screening schedules. Assuming perfect adherence, a baseline policy is extracted. Taking the optimal value from the baseline policy as the predetermined threshold, the optimal number of screening mammograms and screening schedules for various adherence behaviors are obtained from the model.

Result:

The results indicate that depending on women’s adherence behaviors the optimal number of screening mammograms and optimal schedules are significantly different from the current standards, e.g. the American Cancer Society (ACS) guideline. The optimal screening schedule is tight for women with low compliance rate, and has longer intervals between subsequent mammograms for adherent women. However, regardless of adherence behavior the intervals between subsequent mammograms are longer in younger ages, and become shorter as women age.

Conclusion:

Due to variation in women’s adherence behaviors, the efficacy of a “one-size-fits-all” screening policy is questionable. The results of this study can assist physicians/health providers to tailor the screening policies according to their patient adherence behaviors and thereby increasing the probability of detecting cancer before it becomes invasive.