In Reply: Although modeling techniques were not the subject of the Commentary, we agree that experimental, observational, and modeling approaches all have a role to play in evidence development. In the area of modeling, what constitutes quality is challenged not only by limited methodological guidance as to relative strengths and limitations of different models, but also by the fact that key aspects of some models are not transparent.
We agree that use of real-time data from registries to feed decision support tools, including at the point of care, will become increasingly important as electronic systems that can leverage registry data and logical models proliferate. Several registries in many different conditions (such as the American Heart Association Get With the Guidelines registries) have demonstrated a positive effect on quality of care when feedback is provided to clinicians near or at the point of care.1- 3 Like registries, the models that drive prediction also need transparency to allow scrutiny and evaluation.
Dreyer NA, Garner S. Patient Registries, Predictive Models, and Optimal Care—Reply. JAMA. 2009;302(24):2662. doi:10.1001/jama.2009.1890