Communicating with patients and the risks of cancer risk prediction.

Abstract

Abstract Abstract #MS1-4 BACKGROUND
 Patients may ask, “What is my risk of getting cancer?” and they may use cancer risk prediction models that are readily available on the web. Unfortunately, our risk models are less able to predict cancer at the level of the individual patient and our ability to communicate numeric information is limited.
 ASSESSMENT OF RISK MODELS
 Most cancer risk prediction models perform well at the population level. The calibration of risk prediction models is assessed at the population level by comparing the number of patients who the model estimated would develop cancer with the number of patients who actually were diagnosed with cancer.
 Unfortunately, risk prediction models often fall short on their ability to discriminate at the level of the individual patient. The models are less able to separate individuals who are, and are not, going to develop diseases like cancer. Many “low risk” individuals develop cancer, while many “high-risk” individuals do not. Discrimination is the degree to which the estimates from the risk model are consistently higher for individuals who develop cancer, compared to those who do not develop cancer.
 COMMUNICATION
 Communicating information to individuals about their risk of developing cancer is challenging. There is a great deal of fear about cancer in the community and a heightened sense of risk. Much of the discussion about cancer risk is in terms of relative risks, which sound more threatening than absolute risk differences. Use of both positive and negative framing of the risk estimate numbers is helpful. However, the challenge of understanding numbers is not unique to patients; clinicians are also challenged with both understanding and communicating this information.
 CONCLUSION
 Although we have high-tech medical programs and decades of research on cancer, our ability to individually tailor cancer risk prediction tools is sorely lacking. Our current risk prediction models may be appropriate for educating patients about aggregate or population-level cancer risk statistics. However, patients should be counseled that the estimates say little about which specific individuals will develop cancer. We need to develop better risk prediction tools and tools that facilitate basic communication regarding risk and benefit. Both of these challenges may be difficult, but would be extremely worthwhile. Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr MS1-4.