Improving Policy by Pairing Modeling with Sound Data
A hallmark of health care is making difficult decisions in the absence of strong evidence. Health policy makers often issue guidelines without having the benefit of quality data, leading to changes when more information comes to light. Understandably, this can result in frustration and resistance from the public, especially when the recommendations were well-established. For example, there was a backlash to recent changes made in the American Cancer Society’s breast cancer screening guidelines. To combat this, some researchers are turning to mathematical modeling supplemented with comprehensive, real-world data to improve the evidence base for policy decisions. An article published in the Journal of the National Cancer Institute describes their approach.
Mathematical models can predict the impact of interventions on health outcomes. The models’ predictions are based on calculations derived from a variety of sources, including clinical trial data and what’s known of a disease’s natural history. In fact, modeling was used in the development of screening guidelines for cancers of the breast, colon, and lung. Despite the apparent value of models, their accuracy is largely dependent on the quality and comprehensiveness of the data supporting them.
To increase the availability of high-quality data, multi-site research networks are forming to collect large amounts of information on individual behavior related to screening and other matters of health policy. These networks share their data with modeling groups in hopes of improving model predictions. One such network, the Population-based Research Optimizing Screening through Personalized Regimens (PROSPR) consortium, is especially focused on comparing the effectiveness of screening approaches to enhance health policy decision-making.
Though there are challenges inherent in this approach – analyses of the type of data collected by these research networks are prone to certain types of errors – combining high-quality data with sound modeling practices will almost certainly lead to better health policy.
*To learn more about how Hopkins inHealth facilitates acquisition of high-quality data, visit our Core Services page.