Better targeting of medical interventions is ever more important given the aging population , high prices of new technologies and limited budgets . Cost-effectiveness analysis (CEA) assesses the value of interventions, typically by comparing average population health gains to average costs.
Because intervention value differs across individuals, conventional CEA methods should be extended to address patient heterogeneity, when it can be predicted on the basis of pre-treatment characteristics. One way to do so is to incorporate clinical risk prediction modeling and risk stratification into CEAs. However, there are limitations to the accuracy and generalizability of clinical risk prediction models.
In our new article in Medical Decision Making, we use simulation to assess how clinical risk prediction model performance may influence the expected value of individual risk-based versus population-based decision making in different scenarios. We conclude that individualizing treatment decisions using risk may produce substantial value, but also has the potential for net harm. Improvements in model discrimination generally increase the value derived from individualized information; however, when models are miscalibrated, under some circumstances they can lead to worse decisions than the population-wide strategies. These results emphasize the importance of paying special attention to decision context in which models will be deployed and to model calibration, an often neglected aspect of predictive model validation. Caution is needed when transporting risk models to new settings. Whenever possible, risk models should be updated and recalibrated to ensure the accuracy of the risk information they yield.
Olchanski N, Cohen JT, Neumann PJ, Wong JB, Kent D. Understanding the Value of Individualized Information: The Impact of Poor Calibration or Discrimination in Outcome Prediction Models. Medical Decision Making. 2017(in press).