refinement is relative risk. Relative risk is the ratio of incidence for a given disorder in an exposed population to the incidence in an unexposed population. For example, the risk of death by lung cancer might be 15 times higher among smokers than among nonsmokers—a relative risk of 15. Relative risk is obtained in a cohort study and can be approximated in a case-control study.

A second measure linked to risk is attributable risk. Attributable risk is the maximum proportion of cases that would be prevented if an intervention were 100 percent effective in eliminating the risk factor. Attributable risk combines information on relative risk with information on the prevalence of the exposure, in order to help judge which risk factor to target in trying to eliminate the disorder. The formula is p(r − 1)/[p(r − 1) + 1], where p is prevalence of exposure and r is relative risk (Mausner and Kramer, 1985). If the relative risk of lung cancer for smokers versus nonsmokers is 15, as above, and the prevalence of smoking is about 50 percent, then the proportion of cases of lung cancer attributable to smoking, that is, the attributable risk, is approximately 88 percent.

Data on prevalence and on attributable risk are especially germane to research on the prevention of mental disorders. To acquire these data, diverse strategies of research are needed. The prevalence of the disorder is required in order to assess its impact on the population. Prevalence is obtained efficiently from a cross-sectional survey. The attributable risk for a range of risk factors is required in order to select interventions that will have the most powerful effect. Attributable risk probably is most efficiently obtained via the case-control strategy.

An even more recent frontier is the conceptualization of the age of onset for specific disorders. Determination of age of onset is required in order to time the intervention appropriately, that is, before the first incidence of a disorder or problem. Decisions as to when to target a high-risk population can be guided by epidemiological data regarding the range and mean age of onset in a population. Age of onset is best obtained from a longitudinal study of a cohort from the general population. Recognizing the importance of such data, the committee commissioned new analyses of data from the Epidemiologic Catchment Area study. The conceptualizations and methods used in these analyses, and the resulting fresh perspectives they permit, are presented in Chapter 5.


Many scientific areas of study with links to prevention research have their origin in the behavioral and social sciences. Developmental psychopathology is one of these areas. (See Box 4.1 for other illustrations.)

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement