Harvard Report on Cancer Prevention Volume 4: Harvard Cancer Risk Index G.A. Colditz*, K.A. Atwood, K. Emmons, R.R. Monson, W.C. Willett, D. Trichopoulos & D.J. Hunter for the Risk Index Working Group, Harvard Center for Cancer Prevention Channing Laboratory, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School and the Harvard Center for Cancer Prevention, 677 Huntington Avenue, Boston, MA 02115, USA Key words: cancer prevention, risk communication, risk estimation. Abstract Objective: Prediction of cancer risk is a minor component of current health risk appraisals. Perception of individual cancer risk is poor. A Cancer Risk Index was developed to predict individual cancer risk for cancers accounting for 80% of the cancer burden in the United States. Methods: We used group consensus among researchers at the Harvard Medical School and Harvard School of Public Health to identify risk factors as de®nite, probable and possible causes of cancer. Risk points were allocated according to the strength of the causal association and summed. Population average risk of cancer and cumulative 10-year risk was obtained from SEER data. Individual ranking relative to the population average was determined. The risk index was evaluated for validity using colon cancer incidence in prospective cohort data. Results: The Harvard Cancer Risk Index provides a broad classi®cation of cancer risk. Validation against cohort data shows good agreement for colon cancer. Conclusion: The Harvard Cancer Risk Index oers a simple estimation of personal risk of cancer. It may help inform users of the major risk factors for cancer and identify changes in lifestyle that will reduce their risk. It oers the potential for tailored health-promotion messages. Background Prediction rules have become widely used in clinical practice to assist medical decision-making when caring for patients with clinical disease, and to counsel patients regarding the likely course of their disease. Use of such prediction rules for counseling aimed at prevention of chronic illnesses is less well developed. The Framingham Heart Study has served as a basis for development of a prediction rule for future risk of coronary heart disease and for stroke [1, 2], and rules have been developed to predict clinical disease based on results of tests such as exercise stress testing [3]. No parallel prediction rule has been developed for overall risk of cancer, in part because of the many types of cancer that may be considered. Gail et al. have developed a prediction rule for breast cancer that is now used in clinical settings to identify women at high risk, and to counsel them [4, 5]. Dupont and Plummer have prepared a computer program to estimate absolute risk of cancer given known relative risk estimates [6]. Mounting evidence indicates that more than 50% of cancer could be prevented if our current knowledge of risk factors were successfully implemented to reduce risk factor prevalence [7]. We therefore undertook the development of a cancer risk index that might aid physicians in counseling patients about their cancer risk, and may serve to educate the public about the relative importance of cancer risk factors. The need for better informing the public about underlying cancer risk is evident from numerous sources. For example, Black and colleagues showed that women greatly overestimate their own breast cancer risk as well as the value of screening mammography [8]. We limited the Harvard Cancer Risk Index to the leading forms of cancer that account for approximately * Address correspondence to Graham A. Colditz, Channing Lab- oratory, 181 Longwood Avenue, Boston, MA 02115, USA; Ph. 617- 525-2258; Fax 617-525-2008 Cancer Causes and Control 11: 477±488, 2000. 477 Ó 2000 Kluwer Academic Publishers. Printed in the Netherlands.