Confounding and the Analysis of Multiple Variables in Hospital Epidemiology