How Do We Use Multiple Risk Factors? ROC Tree Methods
HELENA CHMURA KRAEMER, Karen Kraemer Lowe, David J. Kupfer · 2005
Abstract The legend of St. Denis, the patron saint of Paris, describes how, after St. Denis was martyred by decapitation, he walked two miles carrying his head in his hands. Centuries later, a wise French woman, Mme. Du Deffand (1697–1780), pointed out: “La distance n’y fait rien, il n’y a que le premier pas qui coute.” Loosely translated: “The distance doesn’t matter—it’s only the first step that counts.” In short, the miracle is that first step. Once that is taken, all he had to do was repeat it. In Chapter 9 we described a “first step”: establishing the potency of a risk factor by acknowledging both the benefit and harm of using a risk factor to predict an outcome. In doing so, we noted that individual risk factors by themselves often are not very potent. Instead, we argued, we must consider multiple risk factors. In Chapter 10 we addressed how multiple risk factors might work together in order to both understand what leads to an outcome and begin to devise ways of preventing (or promoting) it. However, in the end, we had identified which risk factors were important and how they fit together over time but had ignored almost entirely how strong the prediction was.